8GPUサーバー(4th Gen Intel® Xeon® Scalable processors 2CPU)の紹介
前回も8GPUサーバーの紹介をしましたが、今回は4th Gen Intel® Xeon® Scalable processorsを2CPU搭載する8GPUサーバーを紹介します。CPUコア数は最大112、メモリスロット数は32で、128GB DIMMを搭載すると4TB実装可能です。M.2 SSDが2枚、2.5" NVMe SSDを８台、2.5" SATA SSDを8台実装可能です。ネットワークインターフェースは標準で2x 10GbE BaseT with Intel® X710-AT2が付属しています。もちろん専用のIPMI 2.0用1GbE BaseTのインターフェースもあります。シャーシの形状は4Uラックマウント型です。このサイズのサーバーをラックマウントする動画はこちらになります。
これまでと同様に4Uサイズですが、これまでの同型とは違い、2700W/200Vの電源ユニットを4個搭載しています。それに伴い電源ユニットのケーブル用端子形状がIEC-320 C20になっています。付属の電源ケーブルはC19 C20が両端のケーブルです。C13形状のPDUに接続する場合は、別途C19 C14が両端のケーブルが必要です。その場合、2000Wまでしか電源ユニットに電力を供給できなくなりますので、C13形状のPDUを用意して、付属のケーブルを使うのがBestでしょう。電源ユニットは200V用にできており、100Vでの使用は想定されていません。動作もしません。
今回はこのサーバーにRTX4090を8枚、
(base) dl@dl-machine:~$ nvidia-smi
Sat Sep 16 01:21:55 2023
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.104.05 Driver Version: 535.104.05 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 NVIDIA GeForce RTX 4090 On | 00000000:17:00.0 Off | Off |
| 36% 38C P8 19W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 1 NVIDIA GeForce RTX 4090 On | 00000000:2A:00.0 Off | Off |
| 36% 37C P8 12W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 2 NVIDIA GeForce RTX 4090 On | 00000000:3D:00.0 Off | Off |
| 36% 36C P8 12W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 3 NVIDIA GeForce RTX 4090 On | 00000000:63:00.0 Off | Off |
| 35% 35C P8 24W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 4 NVIDIA GeForce RTX 4090 On | 00000000:AB:00.0 Off | Off |
| 36% 38C P8 24W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 5 NVIDIA GeForce RTX 4090 On | 00000000:BD:00.0 Off | Off |
| 34% 37C P8 13W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 6 NVIDIA GeForce RTX 4090 On | 00000000:CF:00.0 Off | Off |
| 35% 38C P8 23W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 7 NVIDIA GeForce RTX 4090 On | 00000000:E1:00.0 Off | Off |
| 34% 37C P8 18W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| 0 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 1 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 2 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 3 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 4 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 5 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 6 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 7 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
+---------------------------------------------------------------------------------------+
(base) dl@dl-machine:~$
CPUは
(base) dl@dl-machine:~/slurm$ head -n 50 /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 143 model name : Intel(R) Xeon(R) Platinum 8462Y+ stepping : 8 microcode : 0x2b0004b1 cpu MHz : 800.078 cache size : 61440 KB physical id : 0 siblings : 32 core id : 0 cpu cores : 32 apicid : 0 initial apicid : 0 fpu : yes fpu_exception : yes cpuid level : 32 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local avx512_bf16 wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid cldemote movdiri movdir64b md_clear pconfig flush_l1d arch_capabilities bugs : spectre_v1 spectre_v2 spec_store_bypass swapgs eibrs_pbrsb bogomips : 5600.00 clflush size : 64 cache_alignment : 64 address sizes : 46 bits physical, 57 bits virtual power management: processor : 1 vendor_id : GenuineIntel cpu family : 6 model : 143 model name : Intel(R) Xeon(R) Platinum 8462Y+ stepping : 8 以下省略
トータルコア数は
base) dl@dl-machine:~/slurm$ nproc 64 (base) dl@dl-machine:~/slurm$
メモリは
(base) dl@dl-machine:~/slurm$ head /proc/meminfo MemTotal: 528054576 kB MemFree: 522731540 kB MemAvailable: 522120056 kB Buffers: 95108 kB Cached: 1956412 kB SwapCached: 0 kB Active: 2091508 kB Inactive: 1118256 kB Active(anon): 1170420 kB Inactive(anon): 33152 kB (base) dl@dl-machine:~/slurm$
というような構成になっています。
今回はgpu burnを使って、8GPUをフル稼働させ、温度がどの程度上昇するかみてみましょう。
ここからリンクをコピーしてgit cloneします。
(base) dl@dl-machine:~$ git clone https://github.com/wilicc/gpu-burn.git Cloning into 'gpu-burn'... remote: Enumerating objects: 224, done. remote: Counting objects: 100% (76/76), done. remote: Compressing objects: 100% (52/52), done. remote: Total 224 (delta 43), reused 42 (delta 24), pack-reused 148 Receiving objects: 100% (224/224), 70.01 KiB | 8.75 MiB/s, done. Resolving deltas: 100% (122/122), done. (base) dl@dl-machine:~$ ls bin Documents go Music singularity tf_cnn_benchmarks cuda-samples Downloads gpu-burn Pictures slurm Videos Desktop examples miniconda3 Public Templates (base) dl@dl-machine:~$
gpu-burnに移動してmakeします。
(base) dl@dl-machine:~$ cd gpu-burn (base) dl@dl-machine:~/gpu-burn$ ls compare.cu Dockerfile gpu_burn-drv.cpp LICENSE Makefile README.md (base) dl@dl-machine:~/gpu-burn$ make (base) dl@dl-machine:~/gpu-burn$ ls compare.cu compare.ptx Dockerfile gpu_burn gpu_burn-drv.cpp gpu_burn-drv.o LICENSE Makefile README.md (base) dl@dl-machine:~/gpu-burn$
実行する前に現在のGPUの温度とFANの回転数を調べておきましょう。
(base) dl@dl-machine:~/gpu-burn$ nvidia-smi
Sat Sep 16 01:59:14 2023
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.104.05 Driver Version: 535.104.05 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 NVIDIA GeForce RTX 4090 On | 00000000:17:00.0 Off | Off |
| 36% 37C P8 16W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 1 NVIDIA GeForce RTX 4090 On | 00000000:2A:00.0 Off | Off |
| 37% 36C P8 11W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 2 NVIDIA GeForce RTX 4090 On | 00000000:3D:00.0 Off | Off |
| 37% 36C P8 12W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 3 NVIDIA GeForce RTX 4090 On | 00000000:63:00.0 Off | Off |
| 36% 35C P8 20W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 4 NVIDIA GeForce RTX 4090 On | 00000000:AB:00.0 Off | Off |
| 37% 37C P8 23W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 5 NVIDIA GeForce RTX 4090 On | 00000000:BD:00.0 Off | Off |
| 34% 36C P8 12W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 6 NVIDIA GeForce RTX 4090 On | 00000000:CF:00.0 Off | Off |
| 36% 37C P8 21W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
| 7 NVIDIA GeForce RTX 4090 On | 00000000:E1:00.0 Off | Off |
| 35% 36C P8 19W / 450W | 13MiB / 24564MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| 0 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 1 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 2 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 3 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 4 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 5 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 6 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
| 7 N/A N/A 3387 G /usr/lib/xorg/Xorg 4MiB |
+---------------------------------------------------------------------------------------+
(base) dl@dl-machine:~/gpu-burn$
GPUの温度は35°C~37°Cになっています。FANの回転数は
[sudo] password for dl: FAN1 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN2 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN3 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN4 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN5 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN6 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN7 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN8 | 3220.000 | RPM | ok | na | 420.000 | na | na | na | na FAN9 | 2240.000 | RPM | ok | na | 420.000 | na | na | na | na FAN10 | 2240.000 | RPM | ok | na | 420.000 | na | na | na | na (base) dl@dl-machine:~/slurm$
になっています。3220回転/分がGPU用FAN、2240回転/分がCPU用FANです。それではストップウォッチを動作させてgpu_burnを実行してみましょう。
2分程度で定常状態に達するのですが、5分経ったところでnvidia-smiでGPUの温度と消費電力を見ると
Tue Sep 19 01:57:04 2023 +---------------------------------------------------------------------------------------+ | NVIDIA-SMI 535.104.05 Driver Version: 535.104.05 CUDA Version: 12.2 | |-----------------------------------------+----------------------+----------------------+ | GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+======================+======================| | 0 NVIDIA GeForce RTX 4090 On | 00000000:17:00.0 Off | Off | | 93% 73C P2 429W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ | 1 NVIDIA GeForce RTX 4090 On | 00000000:2A:00.0 Off | Off | | 95% 74C P2 421W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ | 2 NVIDIA GeForce RTX 4090 On | 00000000:3D:00.0 Off | Off | | 94% 74C P2 409W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ | 3 NVIDIA GeForce RTX 4090 On | 00000000:63:00.0 Off | Off | | 86% 71C P2 421W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ | 4 NVIDIA GeForce RTX 4090 On | 00000000:AB:00.0 Off | Off | | 93% 73C P2 403W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ | 5 NVIDIA GeForce RTX 4090 On | 00000000:BD:00.0 Off | Off | | 83% 69C P2 412W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ | 6 NVIDIA GeForce RTX 4090 On | 00000000:CF:00.0 Off | Off | | 91% 73C P2 413W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ | 7 NVIDIA GeForce RTX 4090 On | 00000000:E1:00.0 Off | Off | | 91% 73C P2 438W / 450W | 21664MiB / 24564MiB | 100% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+
となっており、ファンの回転数は
(base) dl@dl-machine:~/slurm$ sudo ipmitool sensor |grep FAN FAN1 | 8680.000 | RPM | ok | na | 420.000 | na | na | na | na FAN2 | 8820.000 | RPM | ok | na | 420.000 | na | na | na | na FAN3 | 8680.000 | RPM | ok | na | 420.000 | na | na | na | na FAN4 | 8820.000 | RPM | ok | na | 420.000 | na | na | na | na FAN5 | 8400.000 | RPM | ok | na | 420.000 | na | na | na | na FAN6 | 8400.000 | RPM | ok | na | 420.000 | na | na | na | na FAN7 | 8400.000 | RPM | ok | na | 420.000 | na | na | na | na FAN8 | 8400.000 | RPM | ok | na | 420.000 | na | na | na | na FAN9 | 2520.000 | RPM | ok | na | 420.000 | na | na | na | na FAN10 | 2520.000 | RPM | ok | na | 420.000 | na | na | na | na (base) dl@dl-machine:~/slurm$
となっています。
以上、gpu burnの結果でした。
Recent Posts
