Upload
rafe-hill
View
221
Download
0
Tags:
Embed Size (px)
Citation preview
Advanced Memory Hierarchy
**Outline11 Advanced Cache OptimizationsMemory Technology and DRAM optimizationsVirtual MachinesXen VM: Design and PerformanceAMD Opteron Memory HierarchyOpteron Memory Performance vs. Pentium 4Fallacies and PitfallsConclusion
**Why More on Memory Hierarchy?Processor-MemoryPerformance GapGrowing
Chart1
11
1.251.07
1.56251.1449
1.9531251.225043
2.441406251.31079601
3.05175781251.4025517307
3.81469726561.5007303518
5.79833984381.6057814765
8.81347656251.7181861798
13.3964843751.8384592124
20.362656251.9671513573
30.95123752.1048519523
47.0458812.252191589
71.509739122.4098450002
108.69480346242.5785341502
165.21610126282.7590315407
251.12847391952.9521637486
381.71528035773.158815211
580.20722614373.3799322757
881.91498373843.616527535
1340.51077528243.8696844625
2037.57637842924.1405623749
3097.11609521244.4304017411
4707.61646472284.740529863
7155.57702637865.0723669534
8586.69243165445.4274326401
10304.03091798525.8073529249
12364.83710158236.2138676297
14837.80452189876.6488383638
17805.36542627857.1142570492
21366.43851153427.6122550427
Memory
Processor
CPU
DRAM
Year
Performance
Sheet1
CPUDRAM
198011.0025%7%
198111.07
198221.14
198321.23
198421.31
198531.40
198641.5052%
198761.61
198891.72
1989131.84
1990201.97
1991312.10
1992472.25
1993722.41
19941092.58
19951652.76
19962512.95
19973823.16
19985803.38
19998823.62
20001,3413.87
20012,0384.14
20023,0974.43
20034,7084.74
20047,1565.0720%
20058,5875.43
200610,3045.81
200712,3656.21
200814,8386.65
200917,8057.11
201021,3667.61
Sheet1
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
Memory
Processor
CPU
DRAM
Year
Performance
Sheet2
Sheet3
**Review: 6 Basic Cache OptimizationsReducing hit time1.Giving Reads Priority over Writes E.g., Read complete before earlier writes in write bufferAvoiding Address Translation during Cache Indexing Use page offset to index cacheFor physical address matching use tag Reducing Miss Penalty3.Multilevel Caches Reducing Miss Rate4.Larger Block size (Compulsory misses)5.Larger Cache size (Capacity misses)6.Higher Associativity (Conflict misses)
**11 Advanced Cache OptimizationsReducing hit timeSmall and simple cachesWay predictionTrace caches Increasing cache bandwidthPipelined cachesMultibanked cachesNonblocking cachesReducing Miss PenaltyCritical word firstMerging write buffers Reducing Miss RateCompiler optimizations
Reducing miss penalty or miss rate via parallelismHardware prefetchingCompiler prefetching
**1. Fast Hit times via Small and Simple CachesIndex tag memory and then compare takes time Small cache can help hit time since smaller memory takes less time to indexE.g., L1 caches same size for 3 generations of AMD microprocessors: K6, Athlon, and OpteronAlso L2 cache small enough to fit on chip with the processor avoids time penalty of going off chipSimple direct mappingCan overlap tag check with data transmission since no choiceAccess time estimate for 90 nm using CACTI model 4.0Median ratios of access time relative to the direct-mapped caches are 1.32, 1.39, and 1.43 for 2-way, 4-way, and 8-way caches
Chart1
0.55222085650.72597516340.81612716170.8070170238
0.59449834290.74788564710.79700017930.8429381038
0.66866364810.85722505430.89426308210.9571793205
0.76252919751.00795137981.02938690271.0149729883
0.86539354151.23172215781.204178721.2107973918
1.01852166331.64393011511.63867371341.5862286
1.3184589352.11562719942.095071782.0622329802
1-way
2-way
4-way
8-way
Cache size
Access time (ns)
Sheet1
K1-way2-way4-way8-way16
161638416 KB0.550.730.820.811.311.481.46
323276832 KB0.590.750.800.841.261.341.42
646553664 KB0.670.860.890.961.281.341.43
128131072128 KB0.761.011.031.011.09281213961.321.351.33
256262144256 KB0.871.231.201.211.421.391.40
512524288512 KB1.021.641.641.591.611.611.56
102410485761 MB1.322.122.102.061.601.591.56
1.321.391.43
64 byte blocks
.09 microns
Sheet1
1-way
2-way
4-way
8-way
Cache size
Access time (ns)
Sheet2
Sheet3
**2. Fast Hit times via Way PredictionHow to combine fast hit time of Direct Mapped and have the lower conflict misses of 2-way SA cache? Way prediction: keep extra bits in cache to predict the way, or block within the set, of next cache access. Multiplexor is set early to select desired block, only 1 tag comparison performed that clock cycle in parallel with reading the cache data Miss 1st check other blocks for matches in next clock cycle
Accuracy 85%Drawback: CPU pipeline is hard if hit takes 1 or 2 cyclesUsed for instruction caches vs. data cachesHit TimeWay-Miss Hit TimeMiss Penalty
**3. Fast Hit times via Trace Cache (Pentium 4 only; and last time?)Find more instruction level parallelism? How to avoid translation from x86 to microops? Trace cache in Pentium 4Dynamic traces of the executed instructions vs. static sequences of instructions as determined by layout in memoryBuilt-in branch predictorCache the micro-ops vs. x86 instructionsDecode/translate from x86 to micro-ops on trace cache miss+1. better utilize long blocks (dont exit in middle of block, dont enter at label in middle of block)1. complicated address mapping since addresses no longer aligned to power-of-2 multiples of word size-1. instructions may appear multiple times in multiple dynamic traces due to different branch outcomes
**4: Increasing Cache Bandwidth by PipeliningPipeline cache access to maintain bandwidth, but higher latencyInstruction cache access pipeline stages:1: Pentium2: Pentium Pro through Pentium III 4: Pentium 4 greater penalty on mispredicted branches more clock cycles between the issue of the load and the use of the dataAnswer is 3 stages between branch and new instruction fetch and 2 stages between load and use (even though if looked at red insertions that it would be 3 for load and 2 for branch)Reasons:1) Load: TC just does tag check, data available after DS; thus supply the data & forward it, restarting the pipeline on a data cache miss2) EX phase does the address calculation even though just added one phase; presumed reason is that since want fast clock cycle dont want to stick RF phase with reading registers AND testing for zero, so just moved it back on phase
**5. Increasing Cache Bandwidth: Non-Blocking CachesFor pipelined computers that allow out-of-order completion, the processor need not stall on a data cache miss. Processor could continue feteching instructions from instruction cacheNon-blocking cache or lockup-free cache allow data cache to continue to supply cache hits during a missrequires F/E bits on registers for out-of-order executionrequires multi-bank memorieshit under miss reduces the effective miss penalty by working during miss vs. ignoring CPU requestshit under multiple miss or miss under miss may further lower the effective miss penalty by overlapping multiple missesSignificantly increases the complexity of the cache controller as there can be multiple outstanding memory accesses
**6: Increasing Cache Bandwidth via Multiple BanksRather than treat the cache as a single monolithic block, divide into independent banks that can support simultaneous accessesE.g.,T1 (Niagara) L2 has 4 banksBanking works best when accesses naturally spread themselves across banks mapping of addresses to banks affects behavior of memory systemSimple mapping that works well is sequential interleaving Spread block addresses sequentially across banksE,g, if there 4 banks, Bank 0 has all blocks whose address modulo 4 is 0; bank 1 has all blocks whose address modulo 4 is 1;
**7. Reduce Miss Penalty: Early Restart and Critical Word FirstDont wait for full block before restarting CPUEarly restartAs soon as the requested word of the block arrives, send it to the CPU and let the CPU continue executionSpatial locality tend to want next sequential word, so not clear size of benefit of just early restartCritical Word FirstRequest the missed word first from memory and send it to the CPU as soon as it arrives; let the CPU continue execution while filling the rest of the words in the blockLong blocks more popular today Critical Word 1st Widely used
**8. Merging Write Buffer to Reduce Miss PenaltyWrite buffer to allow processor to continue while waiting to write to memoryIf buffer contains modified blocks, the addresses can be checked to see if address of new data matches the address of a valid write buffer entry If so, new data are combined with that entryIncreases block size of write for write-through cache of writes to sequential words, bytes since multiword writes more efficient to memoryThe Sun T1 (Niagara) processor, among many others, uses write merging
**9. Reducing Misses by Compiler OptimizationsMcFarling [1989] showed how to reduce cache misses by 75% on 8KB direct mapped cache, with 4 byte blocks in softwareTechniques for Instructions CacheReorder procedures in memory so as to reduce conflict missesUse profiling to look at conflicts (using tools they have developed)Techniques for Data CacheMerging Arrays: improve spatial locality by single array of compound elements vs. 2 arraysLoop Interchange: change nesting of loops to access data in order stored in memoryLoop Fusion: Combine 2 independent loops that have same looping and some variables overlapBlocking: Improve temporal locality by accessing blocks of data repeatedly vs. going down whole columns or rows
**Merging Arrays Example/* Before: 2 sequential arrays */int val[SIZE];int key[SIZE];
/* After: 1 array of stuctures */struct merge {int val;int key;};struct merge merged_array[SIZE]; Reducing conflicts between val & key; improve spatial locality
**Loop Interchange Example/* Before */for (k = 0; k < 100; k = k+1)for (j = 0; j < 100; j = j+1)for (i = 0; i < 5000; i = i+1)x[i][j] = 2 * x[i][j];/* After */for (k = 0; k < 100; k = k+1)for (i = 0; i < 5000; i = i+1)for (j = 0; j < 100; j = j+1)x[i][j] = 2 * x[i][j]; Sequential accesses instead of striding through memory every 100 words; improved spatial locality
**Loop Fusion Example/* Before */for (i = 0; i < N; i = i+1)for (j = 0; j < N; j = j+1)a[i][j] = 1/b[i][j] * c[i][j];for (i = 0; i < N; i = i+1)for (j = 0; j < N; j = j+1)d[i][j] = a[i][j] + c[i][j];/* After */for (i = 0; i < N; i = i+1)for (j = 0; j < N; j = j+1){a[i][j] = 1/b[i][j] * c[i][j];d[i][j] = a[i][j] + c[i][j];} 2 misses per access to a & c vs. one miss per access; improve spatial locality
**Blocking Example/* Before */for (i = 0; i < N; i = i+1)for (j = 0; j < N; j = j+1){r = 0; for (k = 0; k < N; k = k+1){r = r + y[i][k]*z[k][j];}; x[i][j] = r;};Two Inner Loops:Read all NxN elements of z[]Read N elements of 1 row of y[] repeatedlyWrite N elements of 1 row of x[]Capacity Misses a function of N & Cache Size:2N3 + N2 => (assuming no conflict; otherwise )Idea: compute on BxB submatrix that fits
**Blocking Example/* After */for (jj = 0; jj < N; jj = jj+B)for (kk = 0; kk < N; kk = kk+B)for (i = 0; i < N; i = i+1) for (j = jj; j < min(jj+B-1,N); j = j+1){r = 0; for (k = kk; k < min(kk+B-1,N); k = k+1) {r = r + y[i][k]*z[k][j];}; x[i][j] = x[i][j] + r;}; B called Blocking FactorCapacity Misses from 2N3 + N2 to 2N3/B +N2Conflict Misses Too?
**Reducing Conflict Misses by BlockingEffect of blocking factor on fully associative cache in reducing conflict misses
**Summary of Compiler Optimizations to Reduce Cache Misses
**10. Reducing Misses by Hardware Prefetching of Instructions & DataPrefetching relies on having extra memory bandwidth that can be used without penaltyInstruction PrefetchingTypically, CPU fetches 2 blocks on a miss: the requested block and the next consecutive block. Requested block is placed in instruction cache when it returns, and prefetched block is placed into instruction stream bufferData PrefetchingPentium 4 can prefetch data into L2 cache from up to 8 streams from 8 different 4 KB pages Prefetching invoked if 2 successive L2 cache misses to a page and if distance between those cache blocks is < 256 bytes
Chart1
1.16
1.45
1.18
1.2
1.21
1.26
1.29
1.32
1.4
1.49
1.97
SPECint2000
SPECfp2000
Performance Improvement
Sheet1
Intel Pentium(R) 4 Processor(R) on 90nm Technology, Ronak Singhal (Intel)
HOT CHIPS 16 Archives (2004)
Date August 22-24, 2004 PlaceMemorial Auditorium, Stanford University
gap1.16SPECINT2000
mcf1.45SPECINT2000
fam3d1.18SPECFP2000
wupwise1.20SPECFP2000
galgel1.21SPECFP2000
facerec1.26SPECFP2000
swim1.29SPECFP2000
applu1.32SPECFP2000
lucas1.40SPECFP2000
mgrid1.49SPECFP2000
equake1.97SPECFP2000
applyapplu
?
Sheet1
0
0
0
0
0
0
0
0
0
0
0
SPECint2000
SPECfp2000
Performance Improvement
Sheet2
Sheet3
**11. Reducing Misses by Software Prefetching DataData PrefetchLoad data into register (HP PA-RISC loads)Cache Prefetch: load into cache (MIPS IV, PowerPC, SPARC v. 9)Special prefetching instructions cannot cause faults; a form of speculative execution Issuing Prefetch Instructions takes timeIs cost of prefetch issues < savings in reduced misses?Higher superscalar reduces difficulty of issue bandwidth
**Main Memory BackgroundPerformance of Main Memory: Latency: Cause of Cache Miss PenaltyAccess Time: time between request and word arrivesCycle Time: time between requestsBandwidth: Cause of I/O & Large Block Miss Penalty (L2)Main Memory is DRAM: Dynamic Random Access MemoryDynamic since it needs to be refreshed periodically (8 ms, consumes 1% of time)Addresses divided into 2 halves (Memory as a 2D matrix):RAS or Row Access StrobeCAS or Column Access StrobeCache uses SRAM: Static Random Access MemoryNo refresh (6 transistors/bit vs. 1 transistor) Size: DRAM/SRAM 4-8, Cost: SRAM/DRAM 8-16Cycle time: SRAM/DRAM 8-16
**Main Memory Deep BackgroundOut-of-Core, In-Core, Core Dump?Core memory?Non-volatile, magneticLost to 4 Kbit DRAM (today using 512Mbit DRAM)Access time 750 ns, cycle time 1500-3000 ns
**Quest for DRAM PerformanceFast Page mode Add timing signals that allow repeated accesses to row buffer without another row access timeSuch a buffer comes naturally, as each array will buffer 1024 to 2048 bits for each accessSynchronous DRAM (SDRAM)Add a clock signal to DRAM interface, so that the repeated transfers would not bear overhead to synchronize with DRAM controllerDouble Data Rate (DDR SDRAM)Transfer data on both the rising edge and falling edge of the DRAM clock signal doubling the peak data rateDDR2 lowers power by dropping the voltage from 2.5 to 1.8 volts + offers higher clock rates: up to 400 MHzDDR3 drops to 1.5 volts + higher clock rates: up to 800 MHzImproved Bandwidth, not Latency
* DIMM: dual in-line memory module*DRAM name based on Peak Chip Transfers / SecDIMM name based on Peak DIMM MBytes / Sec
Stan-dardClock Rate (MHz)M transfers / secondDRAM NameMbytes/s/ DIMMDIMM NameDDR133266DDR2662128PC2100DDR150300DDR3002400PC2400DDR200400DDR4003200PC3200DDR2266533DDR2-5334264PC4300DDR2333667DDR2-6675336PC5300DDR2400800DDR2-8006400PC6400DDR35331066DDR3-10668528PC8500DDR36661333DDR3-133310664PC10700DDR38001600DDR3-160012800PC12800
**Need for Error Correction!Motivation:Failures/time proportional to number of bits!As DRAM cells shrink, more vulnerableWent through period in which failure rate was low enough without error correction DRAM banks too large nowServers always corrected memory systemsBasic idea: add redundancy through parity bitsCommon configuration: Random error correctionSEC-DED (single error correct, double error detect)One example: 64 data bits + 8 parity bits (11% overhead)Really want to handle failures of physical components as wellOrganization is multiple DRAMs/DIMM, multiple DIMMsWant to recover from failed DRAM and failed DIMM!Chip kill handle failures width of single DRAM chip
**Introduction to Virtual MachinesVMs developed in late 1960sRemained important in mainframe computing over the yearsLargely ignored in single user computers of 1980s and 1990sRecently regained popularity due toincreasing importance of isolation and security in modern systems, failures in security and reliability of standard operating systems, sharing of a single computer among many unrelated users,and the dramatic increases in raw speed of processors, which makes the overhead of VMs more acceptable
**What is a Virtual Machine (VM)?Broadest definition includes all emulation methods that provide a standard software interface, such as the Java VM(Operating) System Virtual Machines provide a complete system level environment at binary ISAHere assume ISAs always match the native hardware ISAE.g., IBM VM/370, VMware ESX Server, and XenPresent illusion that VM users have entire computer to themselves, including a copy of OSSingle computer runs multiple VMs, and can support a multiple, different OSes On conventional platform, single OS owns all HW resources With a VM, multiple OSes all share HW resourcesUnderlying HW platform is called the host, and its resources are shared among the guest VMs
**Virtual Machine Monitors (VMMs)Virtual machine monitor (VMM) or hypervisor is software that supports VMsVMM determines how to map virtual resources to physical resourcesPhysical resource may be time-shared, partitioned, or emulated in software VMM is much smaller than a traditional OS; isolation portion of a VMM is 10,000 lines of code
**VMM Overhead?Depends on the workloadUser-level processor-bound programs (e.g., SPEC) have zero-virtualization overhead Runs at native speeds since OS rarely invokedI/O-intensive workloads OS-intensive execute many system calls and privileged instructions can result in high virtualization overhead For System VMs, goal of architecture and VMM is to run almost all instructions directly on native hardwareIf I/O-intensive workload is also I/O-bound low processor utilization since waiting for I/O processor virtualization can be hidden low virtualization overhead
**Other Uses of VMsFocus here on protection2 Other commercially important uses of VMsManaging SoftwareVMs provide an abstraction that can run the complete SW stack, even including old OSes like DOSTypical deployment: some VMs running legacy OSes, many running current stable OS release, few testing next OS releaseManaging HardwareVMs allow separate SW stacks to run independently yet share HW, thereby consolidating number of serversSome run each application with compatible version of OS on separate computers, as separation helps dependabilityMigrate running VM to a different computer Either to balance load or to evacuate from failing HW
**Requirements of a Virtual Machine MonitorA VM Monitor Presents a SW interface to guest software, Isolates state of guests from each other, and Protects itself from guest software (including guest OSes)Guest software should behave on a VM exactly as if running on the native HW Except for performance-related behavior or limitations of fixed resources shared by multiple VMsGuest software should not be able to change allocation of real system resources directlyHence, VMM must control everything even though guest VM and OS currently running is temporarily using themAccess to privileged state, Address translation, I/O, Exceptions and Interrupts,
**Requirements of a Virtual Machine MonitorVMM must be at higher privilege level than guest VM, which generally run in user mode Execution of privileged instructions handled by VMME.g., Timer interrupt: VMM suspends currently running guest VM, saves its state, handles interrupt, determine which guest VM to run next, and then load its state Guest VMs that rely on timer interrupt provided with virtual timer and an emulated timer interrupt by VMMRequirements of system virtual machines are same as paged-virtual memory: At least 2 processor modes, system and userPrivileged subset of instructions available only in system mode, trap if executed in user modeAll system resources controllable only via these instructions
**ISA Support for Virtual MachinesIf VMs are planned for during design of ISA, easy to reduce instructions that must be executed by a VMM and how long it takes to emulate themSince VMs have been considered for desktop/PC server apps only recently, most ISAs were created without virtualization in mind, including 80x86 and most RISC architecturesVMM must ensure that guest system only interacts with virtual resources conventional guest OS runs as user mode program on top of VMMIf guest OS attempts to access or modify information related to HW resources via a privileged instruction--for example, reading or writing the page table pointer--it will trap to the VMMIf not, VMM must intercept instruction and support a virtual version of the sensitive information as the guest OS expects (examples soon)
**Impact of VMs on Virtual MemoryVirtualization of virtual memory if each guest OS in every VM manages its own set of page tables?VMM separates real and physical memory Makes real memory a separate, intermediate level between virtual memory and physical memorySome use the terms virtual memory, physical memory, and machine memory to name the 3 levelsGuest OS maps virtual memory to real memory via its page tables, and VMM page tables map real memory to physical memoryVMM maintains a shadow page table that maps directly from the guest virtual address space to the physical address space of HWRather than pay extra level of indirection on every memory accessVMM must trap any attempt by guest OS to change its page table or to access the page table pointer
**ISA Support for VMs & Virtual MemoryIBM 370 architecture added additional level of indirection that is managed by the VMM Guest OS keeps its page tables as before, so the shadow pages are unnecessaryTo virtualize software TLB, VMM manages the real TLB and has a copy of the contents of the TLB of each guest VMAny instruction that accesses the TLB must trapTLBs with Process ID tags support a mix of entries from different VMs and the VMM, thereby avoiding flushing of the TLB on a VM switch
**Impact of I/O on Virtual MemoryMost difficult part of virtualizationIncreasing number of I/O devices attached to the computer Increasing diversity of I/O device typesSharing of a real device among multiple VMs,Supporting the myriad of device drivers that are required, especially if different guest OSes are supported on the same VM systemGive each VM generic versions of each type of I/O device driver, and let VMM to handle real I/OMethod for mapping virtual to physical I/O device depends on the type of device:Disks partitioned by VMM to create virtual disks for guest VMsNetwork interfaces shared between VMs in short time slices, and VMM tracks messages for virtual network addresses to ensure that guest VMs only receive their messages
**Example: Xen VMXen: Open-source System VMM for 80x86 ISA Project started at University of Cambridge, GNU license modelOriginal vision of VM is running unmodified OSSignificant wasted effort just to keep guest OS happyparavirtualization - small modifications to guest OS to simplify virtualization 3 Examples of paravirtualization in Xen:To avoid flushing TLB when invoke VMM, Xen mapped into upper 64 MB of address space of each VM Guest OS allowed to allocate pages, just check that didnt violate protection restrictions To protect the guest OS from user programs in VM, Xen takes advantage of 4 protection levels available in 80x86 Most OSes for 80x86 keep everything at privilege levels 0 or at 3.Xen VMM runs at the highest privilege level (0) Guest OS runs at the next level (1) Applications run at the lowest privilege level (3)
**Xen and I/OTo simplify I/O, privileged VMs assigned to each hardware I/O device: driver domains Xen Jargon: domains = Virtual MachinesDriver domains run physical device drivers, although interrupts still handled by VMM before being sent to appropriate driver domain Regular VMs (guest domains) run simple virtual device drivers that communicate with physical devices drivers in driver domains over a channel to access physical I/O hardware Data sent between guest and driver domains by page remapping
**Xen PerformancePerformance relative to native Linux for Xen for 6 benchmarks from Xen developers
Slide 37: User-level processor-bound programs? I/O-intensive workloads? I/O-Bound I/O-Intensive?
Chart1
1
0.9704797048
0.9186046512
0.9527421237
0.956937799
0.9922779923
Xen/Linux
Performance relative to native Linux
Sheet1
L5671.00Linux (native)Linux (native)Xen (VM)Vmware (VM)Linux (User Mode)
X5671.00Xen (VM)SPEC INT20001.001.000.980.97
V5540.98Vmware (VM)Linux build time1.000.970.790.49
U5500.97Linux (User Mode)OSDB-IR1.000.920.470.38
SPEC INT2000 (score)OSDB-OLTP1.000.950.120.18
L2631.00dbench1.000.960.740.27
X2710.97SPEC WEB991.000.990.290.33
V3340.79Xen/LinuxVMware Workstation 3.2User Mode Linux
U5350.49SPEC INT2000100%1.01.0
Linux build time (s)Linux build time97%0.80.5
L1721.00PostgreSQL Inf. Retrieval92%0.50.4
X1580.92PostgreSQL OLTP95%0.10.2
V800.47dbench96%0.70.3
U650.38SPEC WEB9999%0.30.3
OSDB-IR (tup/s)
L17141.00
X16330.95
V1990.12
U3060.18
OSDB-OLTP (tup/s)
L4181.00
X4000.96
V3100.74
U1110.27
dbench (score)
L5181.00
X5140.99
V1500.29
U1720.33
SPEC WEB99 (score)
Sheet1
00
00
00
00
00
00
Xen/Linux
VMware Workstation 3.2
Performance relative to native Linux
Sheet2
0
0
0
0
0
0
Xen/Linux
Performance relative to native Linux
Sheet3
**Xen Performance, Part IISubsequent study noticed Xen experiments based on 1 Ethernet network interfaces card (NIC), and single NIC was a performance bottleneck
Chart1
942942849
18821878849
24621539849
24461593849
Linux
Xen-privileged driver VM ("driver domain")
Xen-guest VM + driver VM
Number of Network Interface Cards
Receive Throughput (Mbits/sec)
Figure 8 events web server
# commands
# set style data histogram
# set style fill pattern border -1
# set yrange [0:4.5]
# set size 0.6,0.6
# plot 'file' using ($2/$3):xtic(1) title 2, '' using ($3/$3) title 3, '' using ($4/$3) title 4, '' using ($5/$3) title 5
# set xlabel "Profiled Hardware Event"
# set ylabel "Relative costs"
# set terminal postscript eps
# profile for the 1 NIC runs
#profiled parameterLinuxxen-domain0xen-guest0xen-guest1
arbitLinux (1 CPU)Xen-priviledged driver VM (1 CPU)Xen-guest VM + driver VM (1 CPU)Xen-guest VM + driver VM (2 CPUs)
"Instr"13005151473422734815
"L-2"319832001383510470
"I-TLB"423268781196512159
"D-TLB"1119130872462627262
Linux (1 CPU)Xen-priviledged driver VM (1 CPU)Xen-guest VM + driver VM (1 CPU)Xen-guest VM + driver VM (2 CPUs)
"Instr"13005151473422734815
1.22.62.7
"L-2"319832001383510470
1.04.33.3
"I-TLB"423268781196512159
"D-TLB"1119130872462627262
11.722.024.4
all_knot_prof.data for figure 8
Relative to Xen-domain0
Linux (1 CPU)Xen-priviledged driver VM (1 CPU)Xen-guest VM + driver VM (1 CPU)Xen-guest VM + driver VM (2 CPUs)
Intructions0.861.002.262.30
L2 misses1.001.004.323.27
I-TLB misses0.621.001.741.77
D-TLB misses0.091.001.882.08
Figure 8
Figure 8 events web server
0000000000000000
0000000000000000
0000000000000000
0000000000000000
Linux (1 CPU)
Xen-priviledged driver VM (1 CPU)
Xen-guest VM + driver VM (1 CPU)
Xen-guest VM + driver VM (2 CPUs)
Linux (1 CPU)
Xen-priviledged driver VM (1 CPU)
Xen-guest VM + driver VM (1 CPU)
Xen-guest VM + driver VM (2 CPUs)
Linux (1 CPU)
Xen-priviledged driver VM (1 CPU)
Xen-guest VM + driver VM (1 CPU)
Xen-guest VM + driver VM (2 CPUs)
Linux (1 CPU)
Xen-priviledged driver VM (1 CPU)
Xen-guest VM + driver VM (1 CPU)
Xen-guest VM + driver VM (2 CPUs)
Event count relative to Xen-priviledged driver domain
Figure 7 events web server
Request rate (reqs/sec)Throughput (Mbits/sec)
LinuxXen-priviledged driver domainXen-guest0Xen-guest1
10008.138.138.138.13
200016.2716.2716.2716.27
300024.424.424.424.4
400032.5232.5232.5232.52
500040.6640.5240.5240.52
600048.848.845.5148.8
700056.9556.9543.156.95
800065.165.143.165.1
900073.273.145073.14
1000081.4181.4152.581.41
1100089.5389.5353.0389.53
1200097.4697.465397.46
13000105.76105.75105.75
14000113.9113.9113.9
15000122.05122.04107.1
16000130.298.23106.8
17000138.3370.9491.7
18000146.4765.53104.6
19000154.660.26104.51
20000122.7557.02104.5
21000119.2257.14
22000103.0866.33
2300096.8955.83
2400089.7250.53
Figure 7 events web server
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
Linux
Xen-priviledged driver domain
Xen-guest0
Xen-guest1
Figure 3 Rcv Thruput
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
Xen guest VM +privileged driver VM
Xen privileged driver VM only
Lunix
Linux
Xen-priviledged driver domain
Xen-guest0
Request Rate (Reqs/sec)
Throughput (Mbits/sec)
# commands
# set style data histogram
# set style fill pattern border -1
# set yrange [0:3000]
# set size 0.6,0.6
# plot 'file' using 2:xtic(1) title 2, '' using 3 title 3
# set xlabel "Number of NICs"
# set ylabel "Aggregate Throughput (Mb/s)"
# set terminal postscript eps
#number of NICSMb/s(linux)xenXen-domain0
arbitLinuxXen-privileged driver VM ("driver domain")
1942942
218821878
324621539
424461593
LinuxXen-privileged driver VM ("driver domain")Xen-guest VM + driver VM
1942942849
218821878849
324621539849
424461593849
Number of NICs
0
000
000
000
000
Linux
Xen-privileged driver VM ("driver domain")
Xen-guest VM + driver VM
Number of Network Interface Cards
Receive Throughput (Mbits/sec)
MBD00A6E44D.xls
Chart5
0.858585858612.25965537732.2984749455
0.99937514.32343753.271875
0.615295143911.73960453621.76781041
0.085504699311.88171467872.0831359364
Linux
Xen-privileged driver VM only
Xen-guest VM + driver VM
Xen-guest VM + driver VM (2 CPUs)
Event count relative to Xen-priviledged driver domain
Figure 8 events web server
# commands
# set style data histogram
# set style fill pattern border -1
# set yrange [0:4.5]
# set size 0.6,0.6
# plot 'file' using ($2/$3):xtic(1) title 2, '' using ($3/$3) title 3, '' using ($4/$3) title 4, '' using ($5/$3) title 5
# set xlabel "Profiled Hardware Event"
# set ylabel "Relative costs"
# set terminal postscript eps
# profile for the 1 NIC runs
#profiled parameterLinuxxen-domain0xen-guest0xen-guest1
arbitLinuxXen-privileged driver VM onlyXen-guest VM + driver VMXen-guest VM + driver VM (2 CPUs)
"Instr"13005151473422734815
"L-2"319832001383510470
"I-TLB"423268781196512159
"D-TLB"1119130872462627262
LinuxXen-privileged driver VM onlyXen-guest VM + driver VMXen-guest VM + driver VM (2 CPUs)
"Instr"13005151473422734815
"L-2"319832001383510470
"I-TLB"423268781196512159
"D-TLB"1119130872462627262
11.722.024.4
all_knot_prof.data for figure 8
Relative to Xen-domain0
LinuxXen-privileged driver VM onlyXen-guest VM + driver VMXen-guest VM + driver VM (2 CPUs)
Intructions0.861.002.262.30
L2 misses1.001.004.323.27
I-TLB misses0.621.001.741.77
D-TLB misses0.091.001.882.08
Figure 8
Figure 8 events web server
Linux
Xen-privileged driver VM only
Xen-guest VM + driver VM
Event count relative to Xen-priviledged driver domain
Figure 7 events web server
Request rate (reqs/sec)Throughput (Mbits/sec)
LinuxXen-priviledged driver domainXen-guest0Xen-guest1
10008.138.138.138.13
200016.2716.2716.2716.27
300024.424.424.424.4
400032.5232.5232.5232.52
500040.6640.5240.5240.52
600048.848.845.5148.8
700056.9556.9543.156.95
800065.165.143.165.1
900073.273.145073.14
1000081.4181.4152.581.41
1100089.5389.5353.0389.53
1200097.4697.465397.46
13000105.76105.75105.75
14000113.9113.9113.9
15000122.05122.04107.1
16000130.298.23106.8
17000138.3370.9491.7
18000146.4765.53104.6
19000154.660.26104.51
20000122.7557.02104.5
21000119.2257.14
22000103.0866.33
2300096.8955.83
2400089.7250.53
Figure 7 events web server
Linux
Xen-priviledged driver domain
Xen-guest0
Xen-guest1
Figure 3 Rcv Thruput
Xen guest VM+driver VM (1 CPU)
Xen priviledged driver VM (1 CPU)
Lunix(1 CPU)
Xen guest VM+driver VM (2 CPUs)
Linux
Xen-priviledged driver domain
Xen-guest1
Xen-guest0
Request Rate (Reqs/sec)
Throughput (Mbits/sec)
# commands
# set style data histogram
# set style fill pattern border -1
# set yrange [0:3000]
# set size 0.6,0.6
# plot 'file' using 2:xtic(1) title 2, '' using 3 title 3
# set xlabel "Number of NICs"
# set ylabel "Aggregate Throughput (Mb/s)"
# set terminal postscript eps
#number of NICSMb/s(linux)xenXen-domain0
arbitLinuxXen-priviledged driver VM ("driver domain")
1942942
218821878
324621539
424461593
LinuxXen-priviledged driver VM ("driver domain")Xen-guest VM + driver VM
1942942800
218821878930
324621539930
424461593930
Number of NICs
0
Linux
Xen-priviledged driver VM ("driver domain")
Xen-guest VM + driver VM
Number of Network Interface Cards
Receive Throughput (Mbits/sec)
**Xen Performance, Part III> 2X instructions for guest VM + driver VM> 4X L2 cache misses12X 24X Data TLB misses
Chart3
0.858585858612.2596553773
0.99937514.3234375
0.615295143911.7396045362
0.085504699311.8817146787
Linux
Xen-privileged driver VM only
Xen-guest VM + driver VM
Event count relative to Xen-priviledged driver domain
Chart5
0.858585858612.25965537732.2984749455
0.99937514.32343753.271875
0.615295143911.73960453621.76781041
0.085504699311.88171467872.0831359364
Linux
Xen-privileged driver VM only
Xen-guest VM + driver VM
Xen-guest VM + driver VM (2 CPUs)
Event count relative to Xen-priviledged driver domain
Figure 8 events web server
# commands
# set style data histogram
# set style fill pattern border -1
# set yrange [0:4.5]
# set size 0.6,0.6
# plot 'file' using ($2/$3):xtic(1) title 2, '' using ($3/$3) title 3, '' using ($4/$3) title 4, '' using ($5/$3) title 5
# set xlabel "Profiled Hardware Event"
# set ylabel "Relative costs"
# set terminal postscript eps
# profile for the 1 NIC runs
#profiled parameterLinuxxen-domain0xen-guest0xen-guest1
arbitLinuxXen-privileged driver VM onlyXen-guest VM + driver VMXen-guest VM + driver VM (2 CPUs)
"Instr"13005151473422734815
"L-2"319832001383510470
"I-TLB"423268781196512159
"D-TLB"1119130872462627262
LinuxXen-privileged driver VM onlyXen-guest VM + driver VMXen-guest VM + driver VM (2 CPUs)
"Instr"13005151473422734815
"L-2"319832001383510470
"I-TLB"423268781196512159
"D-TLB"1119130872462627262
11.722.024.4
all_knot_prof.data for figure 8
Relative to Xen-domain0
LinuxXen-privileged driver VM onlyXen-guest VM + driver VMXen-guest VM + driver VM (2 CPUs)
Intructions0.861.002.262.30
L2 misses1.001.004.323.27
I-TLB misses0.621.001.741.77
D-TLB misses0.091.001.882.08
Figure 8
Figure 8 events web server
000
000
000
000
Linux
Xen-privileged driver VM only
Xen-guest VM + driver VM
Event count relative to Xen-priviledged driver domain
Figure 7 events web server
Request rate (reqs/sec)Throughput (Mbits/sec)
LinuxXen-priviledged driver domainXen-guest0Xen-guest1
10008.138.138.138.13
200016.2716.2716.2716.27
300024.424.424.424.4
400032.5232.5232.5232.52
500040.6640.5240.5240.52
600048.848.845.5148.8
700056.9556.9543.156.95
800065.165.143.165.1
900073.273.145073.14
1000081.4181.4152.581.41
1100089.5389.5353.0389.53
1200097.4697.465397.46
13000105.76105.75105.75
14000113.9113.9113.9
15000122.05122.04107.1
16000130.298.23106.8
17000138.3370.9491.7
18000146.4765.53104.6
19000154.660.26104.51
20000122.7557.02104.5
21000119.2257.14
22000103.0866.33
2300096.8955.83
2400089.7250.53
Figure 7 events web server
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
Linux
Xen-priviledged driver domain
Xen-guest0
Xen-guest1
Figure 3 Rcv Thruput
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
Xen guest VM+driver VM (1 CPU)
Xen priviledged driver VM (1 CPU)
Lunix(1 CPU)
Xen guest VM+driver VM (2 CPUs)
Linux
Xen-priviledged driver domain
Xen-guest1
Xen-guest0
Request Rate (Reqs/sec)
Throughput (Mbits/sec)
# commands
# set style data histogram
# set style fill pattern border -1
# set yrange [0:3000]
# set size 0.6,0.6
# plot 'file' using 2:xtic(1) title 2, '' using 3 title 3
# set xlabel "Number of NICs"
# set ylabel "Aggregate Throughput (Mb/s)"
# set terminal postscript eps
#number of NICSMb/s(linux)xenXen-domain0
arbitLinuxXen-priviledged driver VM ("driver domain")
1942942
218821878
324621539
424461593
LinuxXen-priviledged driver VM ("driver domain")Xen-guest VM + driver VM
1942942800
218821878930
324621539930
424461593930
Number of NICs
0
000
000
000
000
Linux
Xen-priviledged driver VM ("driver domain")
Xen-guest VM + driver VM
Number of Network Interface Cards
Receive Throughput (Mbits/sec)
**Xen Performance, Part IV> 2X instructions: page remapping and page transfer between driver and guest VMs and due to communication between the 2 VMs over a channel4X L2 cache misses: Linux uses zero-copy network interface that depends on ability of NIC to do DMA from different locations in memory Since Xen does not support gather DMA in its virtual network interface, it cant do true zero-copy in the guest VM12X 24X Data TLB misses: 2 Linux optimizationsSuperpages for part of Linux kernel space, and 4MB pages lowers TLB misses versus using 1024 4 KB pages. Not in XenPTEs marked global are not flushed on a context switch, and Linux uses them for its kernel space. Not in XenFuture Xen may address 2. and 3., but 1. inherent?
**Protection and Instruction Set ArchitectureExample Problem: 80x86 POPF instruction loads flag registers from top of stack in memoryOne such flag is Interrupt Enable (IE)In system mode, POPF changes IE In user mode, POPF simply changes all flags except IE Problem: guest OS runs in user mode inside a VM, so it expects to see changed a IE, but it wontHistorically, IBM mainframe HW and VMM took 3 steps:Reduce cost of processor virtualizationIntel/AMD proposed ISA changes to reduce this costReduce interrupt overhead cost due to virtualizationReduce interrupt cost by steering interrupts to proper VM directly without invoking VMM2. and 3. not yet addressed by Intel/AMD; in the future?
**80x86 VM ChallengesRead control registers in user model that reveal that the guest operating system in running in a virtual machine (such as POPF), and Check protection as required by the segmented architecture but assume that the operating system is running at the highest privilege levelVirtual memory: 80x86 TLBs do not support process ID tags more expensive for VMM and guest OSes to share the TLB each address space change typically requires a TLB flush
**AMD Opteron Memory Hierarchy12-stage integer pipeline yields a maximum clock rate of 2.8 GHz and fastest memory PC3200 DDR SDRAM48-bit virtual and 40-bit physical addressesI and D cache: 64 KB, 2-way set associative, 64-B block, LRUL2 cache: 1 MB, 16-way, 64-B block, pseudo LRUData and L2 caches use write back, write allocate L1 caches are virtually indexed and physically taggedL1 I TLB and L1 D TLB: fully associative, 40 entries 32 entries for 4 KB pages and 8 for 2 MB or 4 MB pages L2 I TLB and L1 D TLB: 4-way, 512 entities of 4 KB pagesMemory controller allows up to 10 cache misses8 from D cache and 2 from I cache
**Opteron Memory Hierarchy PerformanceFor SPEC2000I cache misses per instruction is 0.01% to 0.09% D cache misses per instruction are 1.34% to 1.43% L2 cache misses per instruction are 0.23% to 0.36% Commercial benchmark (TPC-C-like)I cache misses per instruction is 1.83% (100X!)D cache misses per instruction are 1.39% ( same)L2 cache misses per instruction are 0.62% (2X to 3X)How compare to ideal CPI of 0.33?
**CPI breakdown for Integer ProgramsCPI above base attributable to memory 50%L2 cache misses 25% overall (50% memory CPI)Assumes misses are not overlapped with the execution pipeline or with each other, so the pipeline stall portion is a lower bound
Chart2
0.33333333330.106240.2604266667
0.33333333330.096530.2901366667
0.33333333330.003690.4829766667
0.33333333330.151050.3756166667
0.33333333330.137180.3894866667
0.33333333330.296490.2501766667
0.33333333330.565210.1014566667
0.33333333330.302830.3838366667
0.33333333330.346250.6004166667
0.33333333331.204060.2426066667
0.33333333330.931910.5847566667
0.33333333331.251510.9851566667
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
Sheet3
Base CPIPipeline StallMemory CPITotal CPI% CPI-base in Memory
TPC-C like0.330.921.312.5759%
perlbmk0.330.240.130.7035%
crafty0.330.200.190.7249%
eon0.330.480.000.821%
gzip0.330.320.210.8639%
gap0.330.370.150.8629%
vortex0.330.170.380.8869%
bzip20.330.060.611.0091%
gcc0.330.360.331.0248%
parser0.330.540.401.2843%
vpr0.33(0.01)1.461.78101%
twolf0.330.511.011.8566%52%
TPC-C0.330.921.312.5759%53%
CFP2000 Avg0.330.330.491.1560%
sixtrack0.330.260.030.6312%
mesa0.330.350.090.7821%
wupwise0.330.180.310.8363%
mgrid0.330.220.340.8960%
applu0.330.010.620.9798%
facerec0.330.030.711.0796%
galgel0.330.170.571.0777%
apsi0.330.260.581.1769%
ammp0.330.280.581.1968%
fma3d0.330.430.571.3457%
lucas0.330.520.881.7363%
swim0.331.000.551.8835%
equake0.331.210.802.3540%
art0.330.961.733.0364%59%
Sheet3
000
000
000
000
000
000
000
000
000
000
000
000
Base CPI
Pipeline Stall
Memory CPI
CPI
Sheet2
0.33333333330.26238666670.03428
0.33333333330.35187666670.09479
0.33333333330.18241666670.31425
0.33333333330.22061666670.33605
0.33333333330.01378666670.62288
0.33333333330.03115666670.70551
0.33333333330.16735666670.56931
0.33333333330.25520666670.58146
0.33333333330.27824666670.57842
0.33333333330.43484666670.57182
0.33333333330.52112666670.87554
0.33333333331.00030666670.54636
0.33333333331.21468666670.80198
0.33333333330.96347666671.73319
Base CPI
Min Pipeline Stall
Max Memory CPI
CPI
CPI breakdown
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.71
CINT2000 total1.300.9014.273.570.2512.470.001.06
Miss Penalty771607777
Peak CPI0.3333333333
memory CPI
TPC-C like1.310.130.100.990.020.060.000.01
CINT2000 total0.770.010.100.570.000.090.00.01
Base CPI0.33
Memory CPI (no overlap)0.77
Pipeline Stalls (no overlap)0.19
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
Base CPIMemory CPIPipeline StallL2 Memory% Memory L2per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like0.331.310.920.9975%2.5718.3413.896.183.259.000.091.71
Miss Pen771607777
CINT2000 total0.330.770.190.5774%1.300.9014.273.570.2512.470.001.06INT
gzip0.330.210.320.028%0.860.0116.030.100.0111.060.000.09INT
vpr0.331.46(0.01)0.9263%1.780.0223.365.730.0150.520.003.22INT
gcc0.330.330.360.1444%1.021.9419.040.900.794.530.000.19INT
mcf0.3318.20(5.47)16.6191%13.060.02148.90103.820.0150.490.0026.98INT
crafty0.330.190.200.015%0.723.154.050.060.1618.070.000.01INT
parser0.330.400.540.2153%1.280.0814.801.340.0111.560.000.65INT
eon0.330.000.480.00%0.820.060.450.000.010.050.000.00INT
perlbmk0.330.130.240.0754%0.701.362.410.430.933.510.000.31INT
gap0.330.150.370.0960%0.860.764.270.580.053.380.000.33INT
vortex0.330.380.170.1949%0.883.675.861.170.6815.780.001.38INT
bzip20.330.610.060.4778%1.000.0110.572.940.008.170.000.63INT
twolf0.331.010.510.7271%1.850.0826.184.490.0214.790.000.01INT
52%
CFP2000 total0.330.490.330.3674%1.150.0813.432.260.013.700.000.79FP
wupwise0.330.310.180.2785%0.830.006.561.660.000.220.000.17FP
swim0.330.551.000.3259%1.880.0130.872.020.000.590.000.41FP
mgrid0.330.340.220.2264%0.890.0116.541.350.000.350.000.25FP
applu0.330.620.010.5588%0.970.018.483.410.002.420.000.13FP
mesa0.330.090.350.0222%0.780.031.580.130.018.780.000.17FP
galgel0.330.570.170.3867%1.070.0118.632.380.007.620.000.67FP
art0.331.730.961.3276%3.030.0056.968.270.001.200.000.41FP
equake0.330.801.210.5366%2.350.0637.293.300.001.200.000.59FP
facerec0.330.710.030.6389%1.070.019.313.940.001.210.000.20FP
ammp0.330.580.280.3866%1.190.0216.582.370.008.610.003.25FP
lucas0.330.880.520.7080%1.730.0017.354.360.004.800.003.27FP
fma3d0.330.570.430.4885%1.340.2011.843.020.050.360.000.21FP
sixtrack0.330.030.260.0375%0.630.030.530.160.010.660.000.01FP
apsi0.330.580.260.4068%1.170.5013.812.480.0110.370.001.69FP
71%
CINT2000 Avg0.330.190.771.30
mcf0.33(5.47)18.20
Base CPIMin Pipeline StallMax Memory CPITotal CPI
TPC-C like0.330.921.312.57
perlbmk0.330.240.130.70
crafty0.330.200.190.72
eon0.330.480.000.82
gzip0.330.320.210.86
gap0.330.370.150.86
vortex0.330.170.380.88
bzip20.330.060.611.00
gcc0.330.360.331.02
parser0.330.540.401.28
vpr0.33(0.01)1.461.78
twolf0.330.511.011.85
TPC-C0.330.921.312.57
CFP2000 Avg0.330.330.491.15
sixtrack0.330.260.030.63
mesa0.330.350.090.78
wupwise0.330.180.310.83
mgrid0.330.220.340.89
applu0.330.010.620.97
facerec0.330.030.711.07
galgel0.330.170.571.07
apsi0.330.260.581.17
ammp0.330.280.581.19
fma3d0.330.430.571.34
lucas0.330.520.881.73
swim0.331.000.551.88
equake0.331.210.802.35
art0.330.961.733.03
CPI breakdown
0.33333333330.106240.2604266667
0.33333333330.096530.2901366667
0.33333333330.003690.4829766667
0.33333333330.151050.3756166667
0.33333333330.137180.3894866667
0.33333333330.296490.2501766667
0.33333333330.565210.1014566667
0.33333333330.302830.3838366667
0.33333333330.346250.6004166667
0.33333333331.204060.2426066667
0.33333333330.931910.5847566667
0.33333333331.251510.9851566667
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
Sheet1
0.33333333330.034280.2623866667
0.33333333330.094790.3518766667
0.33333333330.314250.1824166667
0.33333333330.336050.2206166667
0.33333333330.622880.0137866667
0.33333333330.705510.0311566667
0.33333333330.569310.1673566667
0.33333333330.581460.2552066667
0.33333333330.578420.2782466667
0.33333333330.571820.4348466667
0.33333333330.875540.5211266667
0.33333333330.546361.0003066667
0.33333333330.801981.2146866667
0.33333333331.733190.9634766667
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
Opteron SPEC CPU2000 Rates
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.71
CINT2000 total1.300.9014.273.570.2512.470.001.06
Miss Penalty771607227
Peak CPI0.3333333333
memory CPI
TPC-C like1.270.130.100.990.020.020.000.01
CINT2000 total0.710.010.100.570.000.020.00.01
Base CPI0.33
Memory CPI (no overlap)0.71
Pipeline Stalls (no overlap)0.26
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
Base CPIMemory CPIPipeline StallL2 Memory% Memory L2per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like0.331.250.990.9979%2.5718.3413.896.183.259.000.091.71
Miss Pen771602277
CINT2000 total0.330.710.260.5780%1.300.9014.273.570.2512.470.001.06INT
perlbmk0.330.110.260.0765%0.701.362.410.430.933.510.000.31INT
crafty0.330.100.290.0110%0.723.154.050.060.1618.070.000.01INT
eon0.330.000.480.00%0.820.060.450.000.010.050.000.00INT
gzip0.330.150.380.0211%0.860.0116.030.100.0111.060.000.09INT
gap0.330.140.390.0968%0.860.764.270.580.053.380.000.33INT
vortex0.330.300.250.1963%0.883.675.861.170.6815.780.001.38INT
bzip20.330.570.100.4783%1.000.0110.572.940.008.170.000.63INT
gcc0.330.300.380.1448%1.021.9419.040.900.794.530.000.19INT
parser0.330.350.600.2162%1.280.0814.801.340.0111.560.000.65INT
vpr0.331.200.240.9276%1.780.0223.365.730.0150.520.003.22INT
twolf0.330.930.580.7277%1.850.0826.184.490.0214.790.000.01INT
mcf0.3317.94(5.22)16.6193%13.060.02148.90103.820.0150.490.0026.98INT
58%
CFP2000 total0.330.470.350.3677%1.150.0813.432.260.013.700.000.79FP
sixtrack0.330.030.270.0383%0.630.030.530.160.010.660.000.01FP
mesa0.330.050.400.0241%0.780.031.580.130.018.780.000.17FP
wupwise0.330.310.180.2785%0.830.006.561.660.000.220.000.17FP
mgrid0.330.330.220.2265%0.890.0116.541.350.000.350.000.25FP
applu0.330.610.030.5589%0.970.018.483.410.002.420.000.13FP
galgel0.330.530.210.3872%1.070.0118.632.380.007.620.000.67FP
facerec0.330.700.040.6390%1.070.019.313.940.001.210.000.20FP
apsi0.330.530.310.4075%1.170.5013.812.480.0110.370.001.69FP
ammp0.330.540.320.3871%1.190.0216.582.370.008.610.003.25FP
fma3d0.330.570.440.4885%1.340.2011.843.020.050.360.000.21FP
lucas0.330.850.550.7082%1.730.0017.354.360.004.800.003.27FP
swim0.330.541.000.3259%1.880.0130.872.020.000.590.000.41FP
equake0.330.801.220.5366%2.350.0637.293.300.001.200.000.59FP
art0.331.730.971.3277%3.030.0056.968.270.001.200.000.41FP
74%
CINT2000 Avg0.330.260.711.30
mcf0.33(5.47)18.20
Base CPIMin Pipeline StallMax Memory CPITotal CPI
TPC-C like0.330.991.252.57
perlbmk0.330.260.110.70
crafty0.330.290.100.72
eon0.330.480.000.82
gzip0.330.380.150.86
gap0.330.390.140.86
vortex0.330.250.300.88
bzip20.330.100.571.00
gcc0.330.380.301.02
parser0.330.600.351.28
vpr0.330.241.201.78
twolf0.330.580.931.85
TPC-C0.330.991.252.57
CFP2000 Avg0.330.350.471.15
sixtrack0.330.260.030.63
mesa0.330.350.090.78
wupwise0.330.180.310.83
mgrid0.330.220.340.89
applu0.330.010.620.97
facerec0.330.030.711.07
galgel0.330.170.571.07
apsi0.330.260.581.17
ammp0.330.280.581.19
fma3d0.330.430.571.34
lucas0.330.520.881.73
swim0.331.000.551.88
equake0.331.210.802.35
art0.330.961.733.03
perlbmk0.330.260.110.70
crafty0.330.290.100.72
eon0.330.480.000.82
gzip0.330.380.150.86
gap0.330.390.140.86
vortex0.330.250.300.88
bzip20.330.100.571.00
gcc0.330.380.301.02
parser0.330.600.351.28
vpr0.330.241.201.78
twolf0.330.580.931.85
TPC-C0.330.991.252.57
sixtrack0.330.260.030.63
mesa0.330.350.090.78
wupwise0.330.180.310.83
mgrid0.330.220.340.89
applu0.330.010.620.97
facerec0.330.030.711.07
galgel0.330.170.571.07
apsi0.330.260.581.17
ammp0.330.280.581.19
fma3d0.330.430.571.34
lucas0.330.520.881.73
swim0.331.000.551.88
equake0.331.210.802.35
art0.330.961.733.03
Opteron SPEC CPU2000 Rates
000
000
000
000
000
000
000
000
000
000
000
000
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
000
000
000
000
000
000
000
000
000
000
000
000
000
000
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 missesIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.711.83%1.39%0.62%0.33%0.90%0.01%0.17%
%%%%%%%
CINT2000 total1.300.9014.273.570.2512.470.001.06INT0.09%1.43%0.36%0.03%1.25%0.00%0.11%
164.gzip0.860.0116.030.100.0111.060.000.09INT0.001%1.603%0.010%0.001%1.106%0.000%0.009%
175.vpr1.780.0223.365.730.0150.520.003.22INT0.002%2.336%0.573%0.001%5.052%0.000%0.322%
176.gcc1.021.9419.040.900.794.530.000.19INT0.194%1.904%0.090%0.079%0.453%0.000%0.019%
181.mcf13.060.02148.90103.820.0150.490.0026.98INT0.002%14.890%10.382%0.001%5.049%0.000%2.698%
186.crafty0.723.154.050.060.1618.070.000.01INT0.315%0.405%0.006%0.016%1.807%0.000%0.001%
197.parser1.280.0814.801.340.0111.560.000.65INT0.008%1.480%0.134%0.001%1.156%0.000%0.065%
252.eon0.820.060.450.000.010.050.000.00INT0.006%0.045%0.000%0.001%0.005%0.000%0.000%
253.perlbmk0.701.362.410.430.933.510.000.31INT0.136%0.241%0.043%0.093%0.351%0.000%0.031%
254.gap0.860.764.270.580.053.380.000.33INT0.076%0.427%0.058%0.005%0.338%0.000%0.033%
255.vortex0.883.675.861.170.6815.780.001.38INT0.367%0.586%0.117%0.068%1.578%0.000%0.138%
256.bzip21.000.0110.572.940.008.170.000.63INT0.001%1.057%0.294%0.000%0.817%0.000%0.063%
300.twolf1.850.0826.184.490.0214.790.000.01INT0.008%2.618%0.449%0.002%1.479%0.000%0.001%
CFP2000 total1.150.0813.432.260.013.700.000.79FP0.01%1.34%0.23%0.00%0.37%0.00%0.08%
168.wupwise0.830.006.561.660.000.220.000.17FP0.000%0.656%0.166%0.000%0.022%0.000%0.017%
171.swim1.880.0130.872.020.000.590.000.41FP0.001%3.087%0.202%0.000%0.059%0.000%0.041%
172.mgrid0.890.0116.541.350.000.350.000.25FP0.001%1.654%0.135%0.000%0.035%0.000%0.025%
173.applu0.970.018.483.410.002.420.000.13FP0.001%0.848%0.341%0.000%0.242%0.000%0.013%
177.mesa0.780.031.580.130.018.780.000.17FP0.003%0.158%0.013%0.001%0.878%0.000%0.017%
178.galgel1.070.0118.632.380.007.620.000.67FP0.001%1.863%0.238%0.000%0.762%0.000%0.067%
179.art3.030.0056.968.270.001.200.000.41FP0.000%5.696%0.827%0.000%0.120%0.000%0.041%
183.equake2.350.0637.293.300.001.200.000.59FP0.006%3.729%0.330%0.000%0.120%0.000%0.059%
187.facerec1.070.019.313.940.001.210.000.20FP0.001%0.931%0.394%0.000%0.121%0.000%0.020%
188.ammp1.190.0216.582.370.008.610.003.25FP0.002%1.658%0.237%0.000%0.861%0.000%0.325%
189.lucas1.730.0017.354.360.004.800.003.27FP0.000%1.735%0.436%0.000%0.480%0.000%0.327%
191.fma3d1.340.2011.843.020.050.360.000.21FP0.020%1.184%0.302%0.005%0.036%0.000%0.021%
200.sixtrack0.630.030.530.160.010.660.000.01FP0.003%0.053%0.016%0.001%0.066%0.000%0.001%
301.apsi1.170.5013.812.480.0110.370.001.69FP0.050%1.381%0.248%0.001%1.037%0.000%0.169%
2.020.41.01.713.00.70.01.6Min0.000%0.045%0.000%0.000%0.005%0.000%0.000%
2.2229.31.02.7325.02.40.02.2Max0.367%14.890%10.382%0.093%5.052%0.000%2.698%
Inverse0.77Icache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
0.87per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
0.38910505842.57Not MCF0.000%0.045%0.000%0.000%0.005%0.000%0.000%
Notes:NOT MCF0.367%5.696%0.827%0.093%5.052%0.000%0.327%
- Data measured on uniprocessor Opteron based systems
- Metrics are averaged over the entire benchmark run
- Only SPEC CPU2000 base results are given. All CPU2000 benchmarks run in 64-bit mode.
- TPC-C like benchmark is run in 32-bit mode.
- Contact: [email protected] 512-602-5581 or 512-576-9485 mobile
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.71
CINT2000 total1.300.9014.273.570.2512.470.001.06INT
164.gzip0.860.0116.030.100.0111.060.000.09INT
175.vpr1.780.0223.365.730.0150.520.003.22INT
176.gcc1.021.9419.040.900.794.530.000.19INT
181.mcf13.060.02148.90103.820.0150.490.0026.98INT
186.crafty0.723.154.050.060.1618.070.000.01INT
197.parser1.280.0814.801.340.0111.560.000.65INT
252.eon0.820.060.450.000.010.050.000.00INT
253.perlbmk0.701.362.410.430.933.510.000.31INT
254.gap0.860.764.270.580.053.380.000.33INT
255.vortex0.883.675.861.170.6815.780.001.38INT
256.bzip21.000.0110.572.940.008.170.000.63INT
300.twolf1.850.0826.184.490.0214.790.000.01INT
CFP2000 total1.150.0813.432.260.013.700.000.79FP
168.wupwise0.830.006.561.660.000.220.000.17FP
171.swim1.880.0130.872.020.000.590.000.41FP
172.mgrid0.890.0116.541.350.000.350.000.25FP
173.applu0.970.018.483.410.002.420.000.13FP
177.mesa0.780.031.580.130.018.780.000.17FP
178.galgel1.070.0118.632.380.007.620.000.67FP
179.art3.030.0056.968.270.001.200.000.41FP
183.equake2.350.0637.293.300.001.200.000.59FP
187.facerec1.070.019.313.940.001.210.000.20FP
188.ammp1.190.0216.582.370.008.610.003.25FP
189.lucas1.730.0017.354.360.004.800.003.27FP
191.fma3d1.340.2011.843.020.050.360.000.21FP
200.sixtrack0.630.030.530.160.010.660.000.01FP
301.apsi1.170.5013.812.480.0110.370.001.69FP
2.020.41.01.713.00.70.01.6
2.2229.31.02.7325.02.40.02.2
Notes:
- Data measured on uniprocessor Opteron based systems
- Metrics are averaged over the entire benchmark run
- Only SPEC CPU2000 base results are given. All CPU2000 benchmarks run in 64-bit mode.
- TPC-C like benchmark is run in 32-bit mode.
- Contact: [email protected] 512-602-5581 or 512-576-9485 mobile
**CPI breakdown for Floating Pt. ProgramsCPI above base attributable to memory 60%L2 cache misses 40% overall (70% memory CPI)Assumes misses are not overlapped with the execution pipeline or with each other, so the pipeline stall portion is a lower bound
Chart4
0.33333333330.034280.2623866667
0.33333333330.094790.3518766667
0.33333333330.314250.1824166667
0.33333333330.336050.2206166667
0.33333333330.622880.0137866667
0.33333333330.705510.0311566667
0.33333333330.569310.1673566667
0.33333333330.581460.2552066667
0.33333333330.578420.2782466667
0.33333333330.571820.4348466667
0.33333333330.875540.5211266667
0.33333333330.546361.0003066667
0.33333333330.801981.2146866667
0.33333333331.733190.9634766667
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
Sheet3
Base CPIPipeline StallMemory CPITotal CPI% CPI-base in Memory
TPC-C like0.330.921.312.5759%
perlbmk0.330.240.130.7035%
crafty0.330.200.190.7249%
eon0.330.480.000.821%
gzip0.330.320.210.8639%
gap0.330.370.150.8629%
vortex0.330.170.380.8869%
bzip20.330.060.611.0091%
gcc0.330.360.331.0248%
parser0.330.540.401.2843%
vpr0.33(0.01)1.461.78101%
twolf0.330.511.011.8566%52%
TPC-C0.330.921.312.5759%53%
CFP2000 Avg0.330.330.491.1560%
sixtrack0.330.260.030.6312%
mesa0.330.350.090.7821%
wupwise0.330.180.310.8363%
mgrid0.330.220.340.8960%
applu0.330.010.620.9798%
facerec0.330.030.711.0796%
galgel0.330.170.571.0777%
apsi0.330.260.581.1769%
ammp0.330.280.581.1968%
fma3d0.330.430.571.3457%
lucas0.330.520.881.7363%
swim0.331.000.551.8835%
equake0.331.210.802.3540%
art0.330.961.733.0364%59%
Sheet3
000
000
000
000
000
000
000
000
000
000
000
000
Base CPI
Pipeline Stall
Memory CPI
CPI
Sheet2
0.33333333330.26238666670.03428
0.33333333330.35187666670.09479
0.33333333330.18241666670.31425
0.33333333330.22061666670.33605
0.33333333330.01378666670.62288
0.33333333330.03115666670.70551
0.33333333330.16735666670.56931
0.33333333330.25520666670.58146
0.33333333330.27824666670.57842
0.33333333330.43484666670.57182
0.33333333330.52112666670.87554
0.33333333331.00030666670.54636
0.33333333331.21468666670.80198
0.33333333330.96347666671.73319
Base CPI
Min Pipeline Stall
Max Memory CPI
CPI
CPI breakdown
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.71
CINT2000 total1.300.9014.273.570.2512.470.001.06
Miss Penalty771607777
Peak CPI0.3333333333
memory CPI
TPC-C like1.310.130.100.990.020.060.000.01
CINT2000 total0.770.010.100.570.000.090.00.01
Base CPI0.33
Memory CPI (no overlap)0.77
Pipeline Stalls (no overlap)0.19
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
Base CPIMemory CPIPipeline StallL2 Memory% Memory L2per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like0.331.310.920.9975%2.5718.3413.896.183.259.000.091.71
Miss Pen771607777
CINT2000 total0.330.770.190.5774%1.300.9014.273.570.2512.470.001.06INT
gzip0.330.210.320.028%0.860.0116.030.100.0111.060.000.09INT
vpr0.331.46(0.01)0.9263%1.780.0223.365.730.0150.520.003.22INT
gcc0.330.330.360.1444%1.021.9419.040.900.794.530.000.19INT
mcf0.3318.20(5.47)16.6191%13.060.02148.90103.820.0150.490.0026.98INT
crafty0.330.190.200.015%0.723.154.050.060.1618.070.000.01INT
parser0.330.400.540.2153%1.280.0814.801.340.0111.560.000.65INT
eon0.330.000.480.00%0.820.060.450.000.010.050.000.00INT
perlbmk0.330.130.240.0754%0.701.362.410.430.933.510.000.31INT
gap0.330.150.370.0960%0.860.764.270.580.053.380.000.33INT
vortex0.330.380.170.1949%0.883.675.861.170.6815.780.001.38INT
bzip20.330.610.060.4778%1.000.0110.572.940.008.170.000.63INT
twolf0.331.010.510.7271%1.850.0826.184.490.0214.790.000.01INT
52%
CFP2000 total0.330.490.330.3674%1.150.0813.432.260.013.700.000.79FP
wupwise0.330.310.180.2785%0.830.006.561.660.000.220.000.17FP
swim0.330.551.000.3259%1.880.0130.872.020.000.590.000.41FP
mgrid0.330.340.220.2264%0.890.0116.541.350.000.350.000.25FP
applu0.330.620.010.5588%0.970.018.483.410.002.420.000.13FP
mesa0.330.090.350.0222%0.780.031.580.130.018.780.000.17FP
galgel0.330.570.170.3867%1.070.0118.632.380.007.620.000.67FP
art0.331.730.961.3276%3.030.0056.968.270.001.200.000.41FP
equake0.330.801.210.5366%2.350.0637.293.300.001.200.000.59FP
facerec0.330.710.030.6389%1.070.019.313.940.001.210.000.20FP
ammp0.330.580.280.3866%1.190.0216.582.370.008.610.003.25FP
lucas0.330.880.520.7080%1.730.0017.354.360.004.800.003.27FP
fma3d0.330.570.430.4885%1.340.2011.843.020.050.360.000.21FP
sixtrack0.330.030.260.0375%0.630.030.530.160.010.660.000.01FP
apsi0.330.580.260.4068%1.170.5013.812.480.0110.370.001.69FP
71%
CINT2000 Avg0.330.190.771.30
mcf0.33(5.47)18.20
Base CPIMin Pipeline StallMax Memory CPITotal CPI
TPC-C like0.330.921.312.57
perlbmk0.330.240.130.70
crafty0.330.200.190.72
eon0.330.480.000.82
gzip0.330.320.210.86
gap0.330.370.150.86
vortex0.330.170.380.88
bzip20.330.060.611.00
gcc0.330.360.331.02
parser0.330.540.401.28
vpr0.33(0.01)1.461.78
twolf0.330.511.011.85
TPC-C0.330.921.312.57
CFP2000 Avg0.330.330.491.15
sixtrack0.330.260.030.63
mesa0.330.350.090.78
wupwise0.330.180.310.83
mgrid0.330.220.340.89
applu0.330.010.620.97
facerec0.330.030.711.07
galgel0.330.170.571.07
apsi0.330.260.581.17
ammp0.330.280.581.19
fma3d0.330.430.571.34
lucas0.330.520.881.73
swim0.331.000.551.88
equake0.331.210.802.35
art0.330.961.733.03
CPI breakdown
0.33333333330.106240.2604266667
0.33333333330.096530.2901366667
0.33333333330.003690.4829766667
0.33333333330.151050.3756166667
0.33333333330.137180.3894866667
0.33333333330.296490.2501766667
0.33333333330.565210.1014566667
0.33333333330.302830.3838366667
0.33333333330.346250.6004166667
0.33333333331.204060.2426066667
0.33333333330.931910.5847566667
0.33333333331.251510.9851566667
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
Sheet1
0.33333333330.034280.2623866667
0.33333333330.094790.3518766667
0.33333333330.314250.1824166667
0.33333333330.336050.2206166667
0.33333333330.622880.0137866667
0.33333333330.705510.0311566667
0.33333333330.569310.1673566667
0.33333333330.581460.2552066667
0.33333333330.578420.2782466667
0.33333333330.571820.4348466667
0.33333333330.875540.5211266667
0.33333333330.546361.0003066667
0.33333333330.801981.2146866667
0.33333333331.733190.9634766667
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
Opteron SPEC CPU2000 Rates
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.71
CINT2000 total1.300.9014.273.570.2512.470.001.06
Miss Penalty771607227
Peak CPI0.3333333333
memory CPI
TPC-C like1.270.130.100.990.020.020.000.01
CINT2000 total0.710.010.100.570.000.020.00.01
Base CPI0.33
Memory CPI (no overlap)0.71
Pipeline Stalls (no overlap)0.26
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
Base CPIMemory CPIPipeline StallL2 Memory% Memory L2per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like0.331.250.990.9979%2.5718.3413.896.183.259.000.091.71
Miss Pen771602277
CINT2000 total0.330.710.260.5780%1.300.9014.273.570.2512.470.001.06INT
perlbmk0.330.110.260.0765%0.701.362.410.430.933.510.000.31INT
crafty0.330.100.290.0110%0.723.154.050.060.1618.070.000.01INT
eon0.330.000.480.00%0.820.060.450.000.010.050.000.00INT
gzip0.330.150.380.0211%0.860.0116.030.100.0111.060.000.09INT
gap0.330.140.390.0968%0.860.764.270.580.053.380.000.33INT
vortex0.330.300.250.1963%0.883.675.861.170.6815.780.001.38INT
bzip20.330.570.100.4783%1.000.0110.572.940.008.170.000.63INT
gcc0.330.300.380.1448%1.021.9419.040.900.794.530.000.19INT
parser0.330.350.600.2162%1.280.0814.801.340.0111.560.000.65INT
vpr0.331.200.240.9276%1.780.0223.365.730.0150.520.003.22INT
twolf0.330.930.580.7277%1.850.0826.184.490.0214.790.000.01INT
mcf0.3317.94(5.22)16.6193%13.060.02148.90103.820.0150.490.0026.98INT
58%
CFP2000 total0.330.470.350.3677%1.150.0813.432.260.013.700.000.79FP
sixtrack0.330.030.270.0383%0.630.030.530.160.010.660.000.01FP
mesa0.330.050.400.0241%0.780.031.580.130.018.780.000.17FP
wupwise0.330.310.180.2785%0.830.006.561.660.000.220.000.17FP
mgrid0.330.330.220.2265%0.890.0116.541.350.000.350.000.25FP
applu0.330.610.030.5589%0.970.018.483.410.002.420.000.13FP
galgel0.330.530.210.3872%1.070.0118.632.380.007.620.000.67FP
facerec0.330.700.040.6390%1.070.019.313.940.001.210.000.20FP
apsi0.330.530.310.4075%1.170.5013.812.480.0110.370.001.69FP
ammp0.330.540.320.3871%1.190.0216.582.370.008.610.003.25FP
fma3d0.330.570.440.4885%1.340.2011.843.020.050.360.000.21FP
lucas0.330.850.550.7082%1.730.0017.354.360.004.800.003.27FP
swim0.330.541.000.3259%1.880.0130.872.020.000.590.000.41FP
equake0.330.801.220.5366%2.350.0637.293.300.001.200.000.59FP
art0.331.730.971.3277%3.030.0056.968.270.001.200.000.41FP
74%
CINT2000 Avg0.330.260.711.30
mcf0.33(5.47)18.20
Base CPIMin Pipeline StallMax Memory CPITotal CPI
TPC-C like0.330.991.252.57
perlbmk0.330.260.110.70
crafty0.330.290.100.72
eon0.330.480.000.82
gzip0.330.380.150.86
gap0.330.390.140.86
vortex0.330.250.300.88
bzip20.330.100.571.00
gcc0.330.380.301.02
parser0.330.600.351.28
vpr0.330.241.201.78
twolf0.330.580.931.85
TPC-C0.330.991.252.57
CFP2000 Avg0.330.350.471.15
sixtrack0.330.260.030.63
mesa0.330.350.090.78
wupwise0.330.180.310.83
mgrid0.330.220.340.89
applu0.330.010.620.97
facerec0.330.030.711.07
galgel0.330.170.571.07
apsi0.330.260.581.17
ammp0.330.280.581.19
fma3d0.330.430.571.34
lucas0.330.520.881.73
swim0.331.000.551.88
equake0.331.210.802.35
art0.330.961.733.03
perlbmk0.330.260.110.70
crafty0.330.290.100.72
eon0.330.480.000.82
gzip0.330.380.150.86
gap0.330.390.140.86
vortex0.330.250.300.88
bzip20.330.100.571.00
gcc0.330.380.301.02
parser0.330.600.351.28
vpr0.330.241.201.78
twolf0.330.580.931.85
TPC-C0.330.991.252.57
sixtrack0.330.260.030.63
mesa0.330.350.090.78
wupwise0.330.180.310.83
mgrid0.330.220.340.89
applu0.330.010.620.97
facerec0.330.030.711.07
galgel0.330.170.571.07
apsi0.330.260.581.17
ammp0.330.280.581.19
fma3d0.330.430.571.34
lucas0.330.520.881.73
swim0.331.000.551.88
equake0.331.210.802.35
art0.330.961.733.03
Opteron SPEC CPU2000 Rates
000
000
000
000
000
000
000
000
000
000
000
000
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
000
000
000
000
000
000
000
000
000
000
000
000
000
000
Base CPI
Max Memory CPI
Min Pipeline Stall
CPI
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
000
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 missesIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.711.83%1.39%0.62%0.33%0.90%0.01%0.17%
%%%%%%%
CINT2000 total1.300.9014.273.570.2512.470.001.06INT0.09%1.43%0.36%0.03%1.25%0.00%0.11%
164.gzip0.860.0116.030.100.0111.060.000.09INT0.001%1.603%0.010%0.001%1.106%0.000%0.009%
175.vpr1.780.0223.365.730.0150.520.003.22INT0.002%2.336%0.573%0.001%5.052%0.000%0.322%
176.gcc1.021.9419.040.900.794.530.000.19INT0.194%1.904%0.090%0.079%0.453%0.000%0.019%
181.mcf13.060.02148.90103.820.0150.490.0026.98INT0.002%14.890%10.382%0.001%5.049%0.000%2.698%
186.crafty0.723.154.050.060.1618.070.000.01INT0.315%0.405%0.006%0.016%1.807%0.000%0.001%
197.parser1.280.0814.801.340.0111.560.000.65INT0.008%1.480%0.134%0.001%1.156%0.000%0.065%
252.eon0.820.060.450.000.010.050.000.00INT0.006%0.045%0.000%0.001%0.005%0.000%0.000%
253.perlbmk0.701.362.410.430.933.510.000.31INT0.136%0.241%0.043%0.093%0.351%0.000%0.031%
254.gap0.860.764.270.580.053.380.000.33INT0.076%0.427%0.058%0.005%0.338%0.000%0.033%
255.vortex0.883.675.861.170.6815.780.001.38INT0.367%0.586%0.117%0.068%1.578%0.000%0.138%
256.bzip21.000.0110.572.940.008.170.000.63INT0.001%1.057%0.294%0.000%0.817%0.000%0.063%
300.twolf1.850.0826.184.490.0214.790.000.01INT0.008%2.618%0.449%0.002%1.479%0.000%0.001%
CFP2000 total1.150.0813.432.260.013.700.000.79FP0.01%1.34%0.23%0.00%0.37%0.00%0.08%
168.wupwise0.830.006.561.660.000.220.000.17FP0.000%0.656%0.166%0.000%0.022%0.000%0.017%
171.swim1.880.0130.872.020.000.590.000.41FP0.001%3.087%0.202%0.000%0.059%0.000%0.041%
172.mgrid0.890.0116.541.350.000.350.000.25FP0.001%1.654%0.135%0.000%0.035%0.000%0.025%
173.applu0.970.018.483.410.002.420.000.13FP0.001%0.848%0.341%0.000%0.242%0.000%0.013%
177.mesa0.780.031.580.130.018.780.000.17FP0.003%0.158%0.013%0.001%0.878%0.000%0.017%
178.galgel1.070.0118.632.380.007.620.000.67FP0.001%1.863%0.238%0.000%0.762%0.000%0.067%
179.art3.030.0056.968.270.001.200.000.41FP0.000%5.696%0.827%0.000%0.120%0.000%0.041%
183.equake2.350.0637.293.300.001.200.000.59FP0.006%3.729%0.330%0.000%0.120%0.000%0.059%
187.facerec1.070.019.313.940.001.210.000.20FP0.001%0.931%0.394%0.000%0.121%0.000%0.020%
188.ammp1.190.0216.582.370.008.610.003.25FP0.002%1.658%0.237%0.000%0.861%0.000%0.325%
189.lucas1.730.0017.354.360.004.800.003.27FP0.000%1.735%0.436%0.000%0.480%0.000%0.327%
191.fma3d1.340.2011.843.020.050.360.000.21FP0.020%1.184%0.302%0.005%0.036%0.000%0.021%
200.sixtrack0.630.030.530.160.010.660.000.01FP0.003%0.053%0.016%0.001%0.066%0.000%0.001%
301.apsi1.170.5013.812.480.0110.370.001.69FP0.050%1.381%0.248%0.001%1.037%0.000%0.169%
2.020.41.01.713.00.70.01.6Min0.000%0.045%0.000%0.000%0.005%0.000%0.000%
2.2229.31.02.7325.02.40.02.2Max0.367%14.890%10.382%0.093%5.052%0.000%2.698%
Inverse0.77Icache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
0.87per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
0.38910505842.57Not MCF0.000%0.045%0.000%0.000%0.005%0.000%0.000%
Notes:NOT MCF0.367%5.696%0.827%0.093%5.052%0.000%0.327%
- Data measured on uniprocessor Opteron based systems
- Metrics are averaged over the entire benchmark run
- Only SPEC CPU2000 base results are given. All CPU2000 benchmarks run in 64-bit mode.
- TPC-C like benchmark is run in 32-bit mode.
- Contact: [email protected] 512-602-5581 or 512-576-9485 mobile
BenchmarkAvg CPIIcache missesDcache missesL2 missesITLB L1 missesDTLB L1 missesITLB L2 missesDTLB L2 misses
per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.per 1k instr.
TPC-C like2.5718.3413.896.183.259.000.091.71
CINT2000 total1.300.9014.273.570.2512.470.001.06INT
164.gzip0.860.0116.030.100.0111.060.000.09INT
175.vpr1.780.0223.365.730.0150.520.003.22INT
176.gcc1.021.9419.040.900.794.530.000.19INT
181.mcf13.060.02148.90103.820.0150.490.0026.98INT
186.crafty0.723.154.050.060.1618.070.000.01INT
197.parser1.280.0814.801.340.0111.560.000.65INT
252.eon0.820.060.450.000.010.050.000.00INT
253.perlbmk0.701.362.410.430.933.510.000.31INT
254.gap0.860.764.270.580.053.380.000.33INT
255.vortex0.883.675.861.170.6815.780.001.38INT
256.bzip21.000.0110.572.940.008.170.000.63INT
300.twolf1.850.0826.184.490.0214.790.000.01INT
CFP2000 total1.150.0813.432.260.013.700.000.79FP
168.wupwise0.830.006.561.660.000.220.000.17FP
171.swim1.880.0130.872.020.000.590.000.41FP
172.mgrid0.890.0116.541.350.000.350.000.25FP
173.applu0.970.018.483.410.002.420.000.13FP
177.mesa0.780.031.580.130.018.780.000.17FP
178.galgel1.070.0118.632.380.007.620.000.67FP
179.art3.030.0056.968.270.001.200.000.41FP
183.equake2.350.0637.293.300.001.200.000.59FP
187.facerec1.070.019.313.940.001.210.000.20FP
188.ammp1.190.0216.582.370.008.610.003.25FP
189.lucas1.730.0017.354.360.004.800.003.27FP
191.fma3d1.340.2011.843.020.050.360.000.21FP
200.sixtrack0.630.030.530.160.010.660.000.01FP
301.apsi1.170.5013.812.480.0110.370.001.69FP
2.020.41.01.713.00.70.01.6
2.2229.31.02.7325.02.40.02.2
Notes:
- Data measured on uniprocessor Opteron based systems
- Metrics are averaged over the entire benchmark run
- Only SPEC CPU2000 base results are given. All CPU2000 benchmarks run in 64-bit mode.
- TPC-C like benchmark is run in 32-bit mode.
- Contact: [email protected] 512-602-5581 or 512-576-9485 mobile
**Pentium 4 vs. Opteron Memory Hierarchy*Clock rate for this comparison in 2005; faster versions existed
CPUPentium 4 (3.2 GHz*)Opteron (2.8 GHz*)Instruction CacheTrace Cache (8K micro-ops)2-way associative, 64 KB, 64B blockData Cache8-way associative, 16 KB, 64B block, inclusive in L22-way associative, 64 KB, 64B block, exclusive to L2L2 cache8-way associative, 2 MB, 128B block16-way associative, 1 MB, 64B blockPrefetch8 streams to L21 stream to L2Memory200 MHz x 64 bits200 MHz x 128 bits
**Misses Per Instruction: Pentium 4 vs. OpteronOpteron betterPentium betterD cache miss: P4 is 2.3X to 3.4X vs. OpteronL2 cache miss: P4 is 0.5X to 1.5X vs. OpteronNote: Same ISA, but not same instruction count2.3X3.4X0.5X1.5X
Chart1
3.49905844833.4651969004
1.77015906670.3413784983
1.62797720580.8163025337
1.2339812360.1755664967
4.68415928980.19326934
4.66914217021.9451292097
5.01586036616.1920687672
1.3099034460.723184898
3.58570534940.4991951275
3.96956736461.6074951627
SPECint2000
SPECfp2000
D cache: P4/Opteron
L2 cache: P4/Opteron
Ratio of MPI: Pentium 4/Opteron
D & L2 MPI, SPEC
Pentium 43.2 GHzOpteron2.8 GHzP4/O
12K16 K2 MBL1/L264 K64 K1 MBL1/L2
I cacheD cacheL2 cacheI cacheD cacheL2 cacheI cacheD cacheL2 cache
164.gzip0.0556.095.610.351620.0116.031.600.101605.03.53.5
175.vpr0.3541.354.141.96210.0223.362.345.73417.61.80.3
176.gcc3.9931.003.100.73421.9419.041.900.90212.11.60.8
181.mcf0.08183.7418.3718.23100.02148.9014.89103.8213.81.20.2
186.crafty5.5518.971.900.011,6363.154.050.410.06681.84.70.2
168.wupwise0.0530.633.063.2390.016.560.661.6649.44.71.9
171.swim0.07154.8415.4812.51120.0130.873.092.02156.85.06.2
172.mgrid0.0221.672.170.98220.0116.541.651.35121.81.30.7
173.applu0.0330.413.041.70180.018.480.853.4122.93.60.5
177.mesa0.516.270.630.21300.031.580.160.131216.94.01.6
MIN0.026.270.630.019.490.011.580.160.061.431.761.230.18
MAX5.55183.7418.3718.231,635.963.15148.9014.89103.82160.3017.635.026.19
I Pent4I OptD Pent4D OptL2 Pent4L2 Opt
gzip0.050.0156.0916.030.350.10
vpr0.350.0241.3523.361.965.73
gcc3.991.9431.0019.040.730.90
mcf0.080.02183.74148.9018.23103.82
crafty5.553.1518.974.050.010.06
wupeise0.050.0130.636.563.231.66
swim0.070.01154.8430.8712.512.02
mgrid0.020.0121.6716.540.981.35
applu0.030.0130.418.481.703.41
mesa0.510.036.271.580.210.13
I cacheD cacheL2 cache
gzip5.03.53.5
vpr17.61.80.3
gcc2.11.60.8
mcf3.81.20.2
crafty1.84.70.2
wupwise9.44.71.9
swim6.85.06.2
mgrid1.81.30.7
applu2.93.60.5
mesa16.94.01.6
D cache: P4/OpteronL2 cache: P4/Opteron
gzip33.5
vpr20.3
gcc20.8
mcf10.2
crafty50.2
wupwise51.9
swim56.2
mgrid10.7
applu40.5
mesa41.6
Median3.50.8
geomean2.760.86
gstdev
SPECratio2000base
P4OpteronRatioRatio Op/P4
gzip106612840.831.20
vpr110312440.891.13
gcc188014311.310.76
mcf196211631.690.59
crafty118318990.621.61
wupwise259728380.921.09
swim264022531.170.85
mgrid146217230.851.18
applu152621220.721.39
mesa140318400.761.31
All GM int158417430.941.10
All GM fp184319761.081.07
GM 5 int1,3871,3821.001.00
GM 5 fp1,8462,1220.871.15
D & L2 MPI, SPEC
000000
000000
000000
000000
000000
000000
000000
000000
000000
000000
I Pent4
I Opt
D Pent4
D Opt
L2 Pent4
L2 Opt
Misses per Intruction
New GeoMean
000
000
000
000
000
000
000
000
000
000
I cache
D cache
L2 cache
GeoMean
00
00
00
00
00
00
00
00
00
00
SPECint2000
SPECfp2000
D cache: P4/Opteron
L2 cache: P4/Opteron
Ratio of MPI: Pentium 4/Opteron
SPEC2000 Pent4
000
000
000
000
000
000
000
000
000
000
D cache: P4/Opteron
L2 cache: P4/Opteron
SPECRatio
Ratio of MPI: Pentium 4/Opteron
SPEC2000 Opt
BenchmarkItanium 2 SPECRatioln(Sample)ln(GM)(ln(S)-ln(GM))**2
gzip0.83(0.19)0.000.04
vpr0.89(0.12)0.000.02
gcc1.310.270.000.07
mcf1.690.520.000.27
Bold are beyondcrafty0.62(0.47)0.000.23
1 St.Dev. Range
N5
1/N0.2Sum ln(S)0.02Sum[(ln(S)-ln(GM))**2]0.62
Product1.0163530546div n0.00Sum[]/n0.12
GM (old)1.00GM (via ln)1.00expSQRT(Sum[]/n)0.35SQRT(Su
GM calculated old wayStDev(Sum(ln(S))0.39St Devexp(SQRT(Sum[]/n))1.42Gstdev
exp(StDev(Sum(ln(S)))1.48Gstdev
GM/Gstdev0.68lowGM/Gstdev0.71low
GM*Gstdev1.49highGM*Gstdev1.43high
GM, Gstdev, Range calculated beforeGM, Gstdev, Range calculated as on web page
BenchmarkItanium 2 SPECRatioln(Sample)ln(GM)(ln(S)-ln(GM))**2
wupwise0.92(0.09)0.000.01
swim1.170.160.000.02
mgrid0.85(0.16)0.000.03
applu0.72(0.33)0.000.11
Bold are beyondmesa0.76(0.27)0.000.08
1 St.Dev. Range
N5
1/N0.2Sum ln(S)(0.70)Sum[(ln(S)-ln(GM))**2]0.25
Product0.4988998701div n(0.14)Sum[]/n0.05
GM (old)0.87GM (via ln)0.87expSQRT(Sum[]/n)0.22SQRT(Su
GM calculated old wayStDev(Sum(ln(S))0.19St Devexp(SQRT(Sum[]/n))1.25Gstdev
exp(StDev(Sum(ln(S)))1.21Gstdev
GM/Gstdev0.72lowGM/Gstdev0.70low
GM*Gstdev1.05highGM*Gstdev1.09high
GM, Gstdev, Range calculated beforeGM, Gstdev, Range calculated as on web page
BenchmarkPentium 4 / Opteron RatiolnD cacheBenchmarkPentium 4 / Opteron RatiolnD cacheBenchmarkPentium 4 / Opteron RatiolnD cache
gzip3.51.25gzip3.51.25
vpr1.80.57vpr1.80.57
gcc1.60.49gcc1.60.49
mcf1.20.21mcf1.20.21
crafty4.71.54crafty4.71.54
wupwise4.71.54wupwise4.71.54
swim5.01.61swim5.01.61
mgrid1.30.27mgrid1.30.27
applu3.61.28applu3.61.28
mesa4.01.38mesa4.01.38
1055
0.125450.224509197810.14Sum ln0.258.28435863264.07Sum ln0.2436.65616481446.08Sum ln
GM2.761.01div nGM2.250.81div nGM3.371.22div n
2.76exp2.25exp3.37exp
0.56St Dev0.56St Dev0.54St Dev
1.76Gstdev1.75Gstdev1.72Gstdev
1.57low1.29low1.96low
4.84high3.95high5.82high
BenchmarkPentium 4 / Opteron RatiolnL2 cacheBenchmarkPentium 4 / Opteron RatiolnL2 cacheBenchmarkPentium 4 / Opteron RatiolnL2 cache
gzip3.51.24gzip3.51.24
vpr0.3(1.07)vpr0.3(1.07)
gcc0.8(0.20)gcc0.8(0.20)
mcf0.2(1.74)mcf0.2(1.74)
crafty0.2(1.64)crafty0.2(1.64)
wupwise1.90.67wupwise1.90.67
swim6.21.82swim6.21.82
mgrid0.7(0.32)mgrid0.7(0.32)
applu0.5(0.69)applu0.5(0.69)
mesa1.60.47mesa1.60.47
1055
0.10.2290200047(1.47)Sum ln0.20.0327657288(3.42)Sum ln0.26.98962035021.94Sum ln
GM0.86(0.15)div nGM0.50(0.68)div nGM1.480.39div n
0.86exp0.50exp1.48exp
1.19St Dev1.24St Dev0.98St Dev
3.30Gstdev3.45Gstdev2.66Gstdev
0low0low1low
3high2high4high
BenchmarkPentium 4 / Opteron RatiolnSPECBenchmarkPentium 4 / Opteron RatiolnL2 cacheBenchmarkPentium 4 / Opteron RatiolnL2 cache
gzip0.83(0.19)gzip0.83(0.19)
vpr0.89(0.12)vpr0.89(0.12)
gcc1.310.27gcc1.310.27
mcf1.690.52mcf1.690.52
crafty0.62(0.47)crafty0.62(0.47)
wupwise0.92(0.09)wupwise0.92(0.09)
swim1.170.16swim1.170.16
mgrid0.85(0.16)mgrid0.85(0.16)
applu0.72(0.33)applu0.72(0.33)
mesa0.76(0.27)mesa0.76(0.27)
1055
0.10.5070584