Hi Mike,
Here's the thread: http://archive.ambermd.org/202010/0081.html
But it looks like the mail archive butchered the formatting. :(
I'll try again with plain text.
Amber 20 Benchmarks
CUDA 11.1
Amber 20 Update 6
AmberTools 20 Update 9
Driver 455.23.04
Throughput ns/day GPU Model
Benchmark RTX2080TI RTX6000 A100 PCIe RTX3080 RTX3090
JAC Production NVE 4fs 970.51 1034.88 1199.22 1101.24 1196.5
JAC Production NPT 4fs 940.61 1004.03 1194.5 1086.21 1157.76
JAC Production NVE 2fs 510.21 540.17 611.08 585.81 632.19
JAC Production NPT 2fs 488.36 515.86 610.09 557.6 595.28
FactorIX Prod NVE 2fs 205.92 217.25 271.36 234.58 264.78
FactorIX Prod NPT 2fs 189.99 206.00 252.87 217.5 248.65
Cellulose Prod NVE 2fs 43.89 47.41 85.23 53.44 63.23
Cellulose Prod NPT 2fs 40.77 45.24 77.98 49.69 58.3
STMV Prod NPT 4fs 25.21 28.49 52.02 32.18 38.65
TRPCage GB 2fs 1151.74 1189.25 1040.61 1332.27 1225.53
Myoglobin GB 2fs 505.93 600.83 661.22 619.67 621.73
Nucleosome GB 2fs 15.22 16.81 29.66 17.72 21.08
Note you will need Amber 20 for things to work. So far in testing all of the test cases are passing on both 3080 and 3090. The validation suite I have also appears to be running fine. My only real worry right now is the power consumption - especially with 3090s:
Sun Sep 27 16:56:56 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 455.23.04 Driver Version: 455.23.04 CUDA Version: 11.1 |
|-------------------------------+----------------------+----------------------+
| 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 GeForce RTX 3090 Off | 00000000:1A:00.0 Off | N/A |
| 67% 71C P2 351W / 350W | 1433MiB / 24268MiB | 99% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
| 1 GeForce RTX 3090 Off | 00000000:68:00.0 Off | N/A |
| 59% 77C P2 349W / 350W | 1433MiB / 24267MiB | 99% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 38372 C ...al/amber20/bin/pmemd.cuda 1431MiB |
| 1 N/A N/A 38373 C ...al/amber20/bin/pmemd.cuda 1431MiB |
+-----------------------------------------------------------------------------+
So Amber manages to easily max out the power limit of the cards. That means 4 of these in a box would be drawing 1.4KW just for the GPUs :-O - that pretty much rules out a 4 x 3090 system running in a US home or office on a 120V x 15amp circuit. You might be okay on a US 120v x 20amp circuit although those can be rare. Europe and UK (and US data centers) will still be okay since they use 220V. This also means that the 3090 is lower performance per watt than the previous 2080TI which is very telling for Moore's law limits and the current state of microprocessor fabrication and doesn't bode well I would think for 4nm and lower. (although some future code improvements from architecture specific optimization might help).
All the best
Ross
> On Oct 9, 2020, at 13:23, Mike Mazanetz <mikem.novadatasolutions.co.uk> wrote:
>
> Hi,
>
> Thanks !
>
> Best,
> mike
> -----Original Message-----
> From: Adrian Roitberg <roitberg.ufl.edu>
> Sent: 09 October 2020 17:45
> To: amber.ambermd.org
> Subject: Re: [AMBER] NVIDIA GeForce RTX 3080 GPUs
>
> Ross walker send a message to the list in the last 2 weeks with A100,
> 3080 and 3090 numbers.
>
> For A100, our own numbers agree with his, so at least two independent sets
> of numbers.
>
> Adrian
>
>
> On 10/9/20 12:41 PM, David Cerutti wrote:
>> [External Email]
>>
>> We have some A100s in a lab or two that have run benchmarks, but we do
>> not yet have official numbers. We do have Amber running on A100 with
>> proper compiler settings, so 3080-Ti and 3090 are not far behind (if
>> we don't have it working already). With no further code changes, the
>> A100 cards were about 30% faster than the reigning V100 for JAC
>> (24,000 atoms), and up to 50% faster for larger systems. Very
>> impressive results, especially considering that the A100 requires that
>> the code uitilize explicit
>> __syncwarp() commands that the earlier cards can do without (these
>> synchronizations amount to 5-15% of the effort if earlier cards are
>> required to use them). This is happening in the context of the cards
>> having about 30% greater fp32 instruction throughput (19.5 TFLOPS)
>> than a V100, so I wouldn't count on new kernel tuning to carry us much
>> further--I suspect the cellulose case, in fact, was benefitting from
>> the better memory bus as much as the higher FLOPS.
>>
>> NVIDIA is once again pushing the envelope, but the real innovations
>> with the A100 are in the fp32 tensor cores which we do not (yet)
>> marshall for our purposes. The fp32 tensor operations are much more
>> suitable for our needs, however, so it's becoming easier to see MD
>> consuming the 150 TFLOPS that the A100's tensor cores provide.
>>
>> One thing I would caution about, especially for people eager to buy
>> Ampere and feed them with existing PCI/Express cables, is that power
>> draw is likely to be a limiting factor. I doubt anyone is thinking
>> about an A100 except in the context of an HGX server, which will have
>> the electricity problem solved out-of-the-box. Gamers are happy with
>> the RTX-3080, saying that it delivers 25% more performance on 10% more
>> juice than the last big thing, the RTX-2080Ti (this is in line with
>> what the A100 should deliver over the RTX-6000 or V100).
>> Specifically, they're talking about a 300-350W power envelope,
>> compared to a 275-300W envelope. The A100 itself is rated up to 400W.
> Take a look at the following specs:
>>
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.nvidia.com_en
>> -2Dus_data-2Dcenter_a100_&d=DwICAg&c=sJ6xIWYx-zLMB3EPkvcnVg&r=dl7Zd5Rz
>> bdvo14I2ndQf4w&m=9eP_ZSN1QqTZLhpL8360yobhLhUtfPM6bVbaaZMn8U4&s=kCaH0MK
>> zjH9iXcUJmEvJrzv1OyfQhL3hebtLytFPEI8&e=
>>
>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__www.nvidia.com_e
>> n-2Dus_data-2Dcenter_a100_&d=DwICAg&c=sJ6xIWYx-zLMB3EPkvcnVg&r=dl7Zd5R
>> zbdvo14I2ndQf4w&m=9eP_ZSN1QqTZLhpL8360yobhLhUtfPM6bVbaaZMn8U4&s=kCaH0M
>> KzjH9iXcUJmEvJrzv1OyfQhL3hebtLytFPEI8&e= >The only difference in the
>> HGX and PCI/E cards? 400W in the former and 250W in the latter, which
>> has to mean that the latter is not going to operate near peak FLOPs
>> for any sustained period of time. my feeling is that, if you hook any
>> Ampere card up to a power supply that wasn't designed to support 350W,
>> you're not going to see that much more performance than you were
>> getting with your RTX-2080Ti or Titan-V. There are some impressive
>> features in the Ampere cards, but the improvement in FLOPs / Watt is
>> incremental and, once again, they are loaded with cores that we and other
> scientific programmers are not yet taking advantage of. The cause for
> optimism is that it is becoming more likely that we will take advantage of
> these innovations with only moderate changes to our algorithms.
>>
>> Dave
>>
>>
>> On Fri, Oct 9, 2020 at 11:37 AM Mike Mazanetz
>> <mikem.novadatasolutions.co.uk>
>> wrote:
>>
>>> Dear All,
>>>
>>>
>>>
>>> I would like to know whether there has been any benchmarking against
>>> the NVIDIA GeForce RTX 3080 GPU cards and whether there is any doubt
>>> whether this would work with
>>>
>>> the latest version of AMBER and previous version, eg AMBER16.
>>>
>>>
>>>
>>> Many thanks,
>>>
>>> mike
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.ambermd.org
>>> _mailman_listinfo_amber&d=DwICAg&c=sJ6xIWYx-zLMB3EPkvcnVg&r=dl7Zd5Rzb
>>> dvo14I2ndQf4w&m=9eP_ZSN1QqTZLhpL8360yobhLhUtfPM6bVbaaZMn8U4&s=_hzlZso
>>> 3bBr4zivF6S2ZHfoVQujIHHf867AxDNvt1HA&e=
>>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.ambermd.org_
>> mailman_listinfo_amber&d=DwICAg&c=sJ6xIWYx-zLMB3EPkvcnVg&r=dl7Zd5Rzbdv
>> o14I2ndQf4w&m=9eP_ZSN1QqTZLhpL8360yobhLhUtfPM6bVbaaZMn8U4&s=_hzlZso3bB
>> r4zivF6S2ZHfoVQujIHHf867AxDNvt1HA&e=
>
> --
> Dr. Adrian E. Roitberg
> V.T. and Louise Jackson Professor in Chemistry Department of Chemistry
> University of Florida roitberg.ufl.edu
> 352-392-6972
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Oct 10 2020 - 07:00:02 PDT
