Project 16921 extremely low ppd

[mgetz]

Normally I wouldn't bring this up, but this set of projects has been abnormally low PPD in comparison to literally every other project. The 169XX series drops down into the 1m PPD range when most other projects regardless of cause range from 2.5m-3m after the recent release of the CUDA core 22 build. Given that the points are supposed to be reflective of the scientific value this to me indicates those projects are low value. I don't mind donating my compute time and power, but I'd rather it be used efficiently particularly right now.

Code: Select all

10:49:19:WU00:FS01:FahCore 0x22 started
10:49:20:WU00:FS01:0x22:*********************** Log Started 2020-10-06T10:49:19Z ***********************
10:49:20:WU00:FS01:0x22:*************************** Core22 Folding@home Core ***************************
10:49:20:WU00:FS01:0x22:       Core: Core22
10:49:20:WU00:FS01:0x22:       Type: 0x22
10:49:20:WU00:FS01:0x22:    Version: 0.0.13
10:49:20:WU00:FS01:0x22:     Author: Joseph Coffland <[email protected]>
10:49:20:WU00:FS01:0x22:  Copyright: 2020 foldingathome.org
10:49:20:WU00:FS01:0x22:   Homepage: https://foldingathome.org/
10:49:20:WU00:FS01:0x22:       Date: Sep 19 2020
10:49:20:WU00:FS01:0x22:       Time: 02:35:58
10:49:20:WU00:FS01:0x22:   Revision: 571cf95de6de2c592c7c3ed48fcfb2e33e9ea7d3
10:49:20:WU00:FS01:0x22:     Branch: core22-0.0.13
10:49:20:WU00:FS01:0x22:   Compiler: Visual C++ 2015
10:49:20:WU00:FS01:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
10:49:20:WU00:FS01:0x22:             -DOPENMM_GIT_HASH="\"189320d0\""
10:49:20:WU00:FS01:0x22:   Platform: win32 10
10:49:20:WU00:FS01:0x22:       Bits: 64
10:49:20:WU00:FS01:0x22:       Mode: Release
10:49:20:WU00:FS01:0x22:Maintainers: John Chodera <[email protected]> and Peter Eastman
10:49:20:WU00:FS01:0x22:             <[email protected]>
10:49:20:WU00:FS01:0x22:       Args: -dir 00 -suffix 01 -version 706 -lifeline 19392 -checkpoint 15
10:49:20:WU00:FS01:0x22:             -gpu-vendor nvidia -opencl-platform 0 -opencl-device 0 -cuda-device
10:49:20:WU00:FS01:0x22:             0 -gpu 0
10:49:20:WU00:FS01:0x22:************************************ libFAH ************************************
10:49:20:WU00:FS01:0x22:       Date: Sep 7 2020
10:49:20:WU00:FS01:0x22:       Time: 19:09:56
10:49:20:WU00:FS01:0x22:   Revision: 44301ed97b996b63fe736bb8073f22209cb2b603
10:49:20:WU00:FS01:0x22:     Branch: HEAD
10:49:20:WU00:FS01:0x22:   Compiler: Visual C++ 2015
10:49:20:WU00:FS01:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
10:49:20:WU00:FS01:0x22:   Platform: win32 10
10:49:20:WU00:FS01:0x22:       Bits: 64
10:49:20:WU00:FS01:0x22:       Mode: Release
10:49:20:WU00:FS01:0x22:************************************ CBang *************************************
10:49:20:WU00:FS01:0x22:       Date: Sep 7 2020
10:49:20:WU00:FS01:0x22:       Time: 19:08:30
10:49:20:WU00:FS01:0x22:   Revision: 33fcfc2b3ed2195a423606a264718e31e6b3903f
10:49:20:WU00:FS01:0x22:     Branch: HEAD
10:49:20:WU00:FS01:0x22:   Compiler: Visual C++ 2015
10:49:20:WU00:FS01:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /O2 /Ob3 /Zc:throwingNew /MT
10:49:20:WU00:FS01:0x22:   Platform: win32 10
10:49:20:WU00:FS01:0x22:       Bits: 64
10:49:20:WU00:FS01:0x22:       Mode: Release
10:49:20:WU00:FS01:0x22:************************************ System ************************************
10:49:20:WU00:FS01:0x22:        CPU: AMD Ryzen 7 3800X 8-Core Processor
10:49:20:WU00:FS01:0x22:     CPU ID: AuthenticAMD Family 23 Model 113 Stepping 0
10:49:20:WU00:FS01:0x22:       CPUs: 16
10:49:20:WU00:FS01:0x22:     Memory: 15.92GiB
10:49:20:WU00:FS01:0x22:Free Memory: 11.06GiB
10:49:20:WU00:FS01:0x22:    Threads: WINDOWS_THREADS
10:49:20:WU00:FS01:0x22: OS Version: 6.2
10:49:20:WU00:FS01:0x22:Has Battery: false
10:49:20:WU00:FS01:0x22: On Battery: false
10:49:20:WU00:FS01:0x22: UTC Offset: -6
10:49:20:WU00:FS01:0x22:        PID: 8120
10:49:20:WU00:FS01:0x22:        CWD: C:\Users\<censored>\AppData\Roaming\FAHClient\work
10:49:20:WU00:FS01:0x22:************************************ OpenMM ************************************
10:49:20:WU00:FS01:0x22:   Revision: 189320d0
10:49:20:WU00:FS01:0x22:********************************************************************************
10:49:20:WU00:FS01:0x22:Project: 16921 (Run 50, Clone 36, Gen 51)
10:49:20:WU00:FS01:0x22:Unit: 0x000000360002894c5f5bf353ce00a10a
10:49:20:WU00:FS01:0x22:Reading tar file core.xml
10:49:20:WU00:FS01:0x22:Reading tar file integrator.xml
10:49:20:WU00:FS01:0x22:Reading tar file state.xml
10:49:20:WU00:FS01:0x22:Reading tar file system.xml
10:49:20:WU00:FS01:0x22:Digital signatures verified
10:49:20:WU00:FS01:0x22:Folding@home GPU Core22 Folding@home Core
10:49:20:WU00:FS01:0x22:Version 0.0.13
10:49:20:WU00:FS01:0x22:  Checkpoint write interval: 250000 steps (5%) [20 total]
10:49:20:WU00:FS01:0x22:  JSON viewer frame write interval: 50000 steps (1%) [100 total]
10:49:20:WU00:FS01:0x22:  XTC frame write interval: 250000 steps (5%) [20 total]
10:49:20:WU00:FS01:0x22:  Global context and integrator variables write interval: disabled
10:49:20:WU00:FS01:0x22:There are 4 platforms available.
10:49:20:WU00:FS01:0x22:Platform 0: Reference
10:49:20:WU00:FS01:0x22:Platform 1: CPU
10:49:20:WU00:FS01:0x22:Platform 2: OpenCL
10:49:20:WU00:FS01:0x22:  opencl-device 0 specified
10:49:20:WU00:FS01:0x22:Platform 3: CUDA
10:49:20:WU00:FS01:0x22:  cuda-device 0 specified
10:49:21:WU00:FS01:0x22:Attempting to create CUDA context:
10:49:21:WU00:FS01:0x22:  Configuring platform CUDA
10:49:25:WU00:FS01:0x22:  Using CUDA and gpu 0
10:49:25:WU00:FS01:0x22:Completed 0 out of 5000000 steps (0%)
10:49:25:WU00:FS01:0x22:Checkpoint completed at step 0
10:49:51:WU00:FS01:0x22:Completed 50000 out of 5000000 steps (1%)
10:50:17:WU00:FS01:0x22:Completed 100000 out of 5000000 steps (2%)
10:50:43:WU00:FS01:0x22:Completed 150000 out of 5000000 steps (3%)
10:51:09:WU00:FS01:0x22:Completed 200000 out of 5000000 steps (4%)
10:51:33:WU00:FS01:0x22:Completed 250000 out of 5000000 steps (5%)
10:51:33:WU00:FS01:0x22:Checkpoint completed at step 250000
10:51:57:WU00:FS01:0x22:Completed 300000 out of 5000000 steps (6%)
10:52:21:WU00:FS01:0x22:Completed 350000 out of 5000000 steps (7%)
10:52:45:WU00:FS01:0x22:Completed 400000 out of 5000000 steps (8%)
10:53:09:WU00:FS01:0x22:Completed 450000 out of 5000000 steps (9%)
10:53:32:WU00:FS01:0x22:Completed 500000 out of 5000000 steps (10%)
10:53:32:WU00:FS01:0x22:Checkpoint completed at step 500000
10:53:56:WU00:FS01:0x22:Completed 550000 out of 5000000 steps (11%)
10:54:20:WU00:FS01:0x22:Completed 600000 out of 5000000 steps (12%)
10:54:44:WU00:FS01:0x22:Completed 650000 out of 5000000 steps (13%)
10:55:09:WU00:FS01:0x22:Completed 700000 out of 5000000 steps (14%)
10:55:33:WU00:FS01:0x22:Completed 750000 out of 5000000 steps (15%)
10:55:34:WU00:FS01:0x22:Checkpoint completed at step 750000
10:55:58:WU00:FS01:0x22:Completed 800000 out of 5000000 steps (16%)
10:56:22:WU00:FS01:0x22:Completed 850000 out of 5000000 steps (17%)
10:56:46:WU00:FS01:0x22:Completed 900000 out of 5000000 steps (18%)
10:57:09:WU00:FS01:0x22:Completed 950000 out of 5000000 steps (19%)
10:57:33:WU00:FS01:0x22:Completed 1000000 out of 5000000 steps (20%)
10:57:33:WU00:FS01:0x22:Checkpoint completed at step 1000000
10:57:57:WU00:FS01:0x22:Completed 1050000 out of 5000000 steps (21%)
10:58:21:WU00:FS01:0x22:Completed 1100000 out of 5000000 steps (22%)
10:58:45:WU00:FS01:0x22:Completed 1150000 out of 5000000 steps (23%)
10:59:09:WU00:FS01:0x22:Completed 1200000 out of 5000000 steps (24%)
10:59:32:WU00:FS01:0x22:Completed 1250000 out of 5000000 steps (25%)
10:59:32:WU00:FS01:0x22:Checkpoint completed at step 1250000
10:59:56:WU00:FS01:0x22:Completed 1300000 out of 5000000 steps (26%)
11:00:20:WU00:FS01:0x22:Completed 1350000 out of 5000000 steps (27%)
11:00:44:WU00:FS01:0x22:Completed 1400000 out of 5000000 steps (28%)
11:01:08:WU00:FS01:0x22:Completed 1450000 out of 5000000 steps (29%)
11:01:32:WU00:FS01:0x22:Completed 1500000 out of 5000000 steps (30%)
11:01:32:WU00:FS01:0x22:Checkpoint completed at step 1500000
11:01:56:WU00:FS01:0x22:Completed 1550000 out of 5000000 steps (31%)
11:02:21:WU00:FS01:0x22:Completed 1600000 out of 5000000 steps (32%)
11:02:45:WU00:FS01:0x22:Completed 1650000 out of 5000000 steps (33%)
11:03:09:WU00:FS01:0x22:Completed 1700000 out of 5000000 steps (34%)
11:03:33:WU00:FS01:0x22:Completed 1750000 out of 5000000 steps (35%)
11:03:33:WU00:FS01:0x22:Checkpoint completed at step 1750000
11:03:57:WU00:FS01:0x22:Completed 1800000 out of 5000000 steps (36%)
11:04:22:WU00:FS01:0x22:Completed 1850000 out of 5000000 steps (37%)
11:04:46:WU00:FS01:0x22:Completed 1900000 out of 5000000 steps (38%)
11:05:10:WU00:FS01:0x22:Completed 1950000 out of 5000000 steps (39%)
11:05:34:WU00:FS01:0x22:Completed 2000000 out of 5000000 steps (40%)
11:05:34:WU00:FS01:0x22:Checkpoint completed at step 2000000
11:05:58:WU00:FS01:0x22:Completed 2050000 out of 5000000 steps (41%)
11:06:21:WU00:FS01:0x22:Completed 2100000 out of 5000000 steps (42%)
11:06:45:WU00:FS01:0x22:Completed 2150000 out of 5000000 steps (43%)
11:07:09:WU00:FS01:0x22:Completed 2200000 out of 5000000 steps (44%)
11:07:33:WU00:FS01:0x22:Completed 2250000 out of 5000000 steps (45%)
11:07:33:WU00:FS01:0x22:Checkpoint completed at step 2250000
11:07:58:WU00:FS01:0x22:Completed 2300000 out of 5000000 steps (46%)
11:08:22:WU00:FS01:0x22:Completed 2350000 out of 5000000 steps (47%)
11:08:45:WU00:FS01:0x22:Completed 2400000 out of 5000000 steps (48%)
11:09:09:WU00:FS01:0x22:Completed 2450000 out of 5000000 steps (49%)
11:09:34:WU00:FS01:0x22:Completed 2500000 out of 5000000 steps (50%)
11:09:34:WU00:FS01:0x22:Checkpoint completed at step 2500000
11:09:58:WU00:FS01:0x22:Completed 2550000 out of 5000000 steps (51%)
11:10:22:WU00:FS01:0x22:Completed 2600000 out of 5000000 steps (52%)
11:10:46:WU00:FS01:0x22:Completed 2650000 out of 5000000 steps (53%)
11:11:10:WU00:FS01:0x22:Completed 2700000 out of 5000000 steps (54%)
11:11:34:WU00:FS01:0x22:Completed 2750000 out of 5000000 steps (55%)
11:11:34:WU00:FS01:0x22:Checkpoint completed at step 2750000
11:11:58:WU00:FS01:0x22:Completed 2800000 out of 5000000 steps (56%)
11:12:22:WU00:FS01:0x22:Completed 2850000 out of 5000000 steps (57%)
11:12:47:WU00:FS01:0x22:Completed 2900000 out of 5000000 steps (58%)
11:13:11:WU00:FS01:0x22:Completed 2950000 out of 5000000 steps (59%)
11:13:35:WU00:FS01:0x22:Completed 3000000 out of 5000000 steps (60%)
11:13:35:WU00:FS01:0x22:Checkpoint completed at step 3000000
11:13:59:WU00:FS01:0x22:Completed 3050000 out of 5000000 steps (61%)
11:14:23:WU00:FS01:0x22:Completed 3100000 out of 5000000 steps (62%)
11:14:47:WU00:FS01:0x22:Completed 3150000 out of 5000000 steps (63%)
11:15:11:WU00:FS01:0x22:Completed 3200000 out of 5000000 steps (64%)
11:15:35:WU00:FS01:0x22:Completed 3250000 out of 5000000 steps (65%)
11:15:35:WU00:FS01:0x22:Checkpoint completed at step 3250000
11:15:59:WU00:FS01:0x22:Completed 3300000 out of 5000000 steps (66%)
11:16:23:WU00:FS01:0x22:Completed 3350000 out of 5000000 steps (67%)
11:16:47:WU00:FS01:0x22:Completed 3400000 out of 5000000 steps (68%)
11:17:11:WU00:FS01:0x22:Completed 3450000 out of 5000000 steps (69%)
11:17:35:WU00:FS01:0x22:Completed 3500000 out of 5000000 steps (70%)
11:17:35:WU00:FS01:0x22:Checkpoint completed at step 3500000
11:17:59:WU00:FS01:0x22:Completed 3550000 out of 5000000 steps (71%)
11:18:24:WU00:FS01:0x22:Completed 3600000 out of 5000000 steps (72%)
11:18:47:WU00:FS01:0x22:Completed 3650000 out of 5000000 steps (73%)
11:19:11:WU00:FS01:0x22:Completed 3700000 out of 5000000 steps (74%)
11:19:35:WU00:FS01:0x22:Completed 3750000 out of 5000000 steps (75%)
11:19:35:WU00:FS01:0x22:Checkpoint completed at step 3750000
11:19:59:WU00:FS01:0x22:Completed 3800000 out of 5000000 steps (76%)
11:20:23:WU00:FS01:0x22:Completed 3850000 out of 5000000 steps (77%)
11:20:47:WU00:FS01:0x22:Completed 3900000 out of 5000000 steps (78%)
11:21:11:WU00:FS01:0x22:Completed 3950000 out of 5000000 steps (79%)
11:21:35:WU00:FS01:0x22:Completed 4000000 out of 5000000 steps (80%)
11:21:35:WU00:FS01:0x22:Checkpoint completed at step 4000000
11:21:59:WU00:FS01:0x22:Completed 4050000 out of 5000000 steps (81%)
11:22:23:WU00:FS01:0x22:Completed 4100000 out of 5000000 steps (82%)
11:22:47:WU00:FS01:0x22:Completed 4150000 out of 5000000 steps (83%)
11:23:11:WU00:FS01:0x22:Completed 4200000 out of 5000000 steps (84%)
11:23:35:WU00:FS01:0x22:Completed 4250000 out of 5000000 steps (85%)
11:23:35:WU00:FS01:0x22:Checkpoint completed at step 4250000
11:23:58:WU00:FS01:0x22:Completed 4300000 out of 5000000 steps (86%)
11:24:23:WU00:FS01:0x22:Completed 4350000 out of 5000000 steps (87%)
11:24:47:WU00:FS01:0x22:Completed 4400000 out of 5000000 steps (88%)
11:25:11:WU00:FS01:0x22:Completed 4450000 out of 5000000 steps (89%)
11:25:35:WU00:FS01:0x22:Completed 4500000 out of 5000000 steps (90%)
11:25:35:WU00:FS01:0x22:Checkpoint completed at step 4500000
11:25:59:WU00:FS01:0x22:Completed 4550000 out of 5000000 steps (91%)
11:26:23:WU00:FS01:0x22:Completed 4600000 out of 5000000 steps (92%)
11:26:46:WU00:FS01:0x22:Completed 4650000 out of 5000000 steps (93%)
11:27:10:WU00:FS01:0x22:Completed 4700000 out of 5000000 steps (94%)
11:27:34:WU00:FS01:0x22:Completed 4750000 out of 5000000 steps (95%)
11:27:34:WU00:FS01:0x22:Checkpoint completed at step 4750000
11:27:58:WU00:FS01:0x22:Completed 4800000 out of 5000000 steps (96%)
11:28:22:WU00:FS01:0x22:Completed 4850000 out of 5000000 steps (97%)
11:28:46:WU00:FS01:0x22:Completed 4900000 out of 5000000 steps (98%)
11:29:10:WU00:FS01:0x22:Completed 4950000 out of 5000000 steps (99%)
11:29:33:WU00:FS01:0x22:Completed 5000000 out of 5000000 steps (100%)
11:29:33:WU00:FS01:0x22:Average performance: 363.025 ns/day

13-10-2020: edited title to reflect this is a single project issue not a series issue.

[PantherX]

Can you please tell us what the GPU model is? Based on the log above, it seems to be a Nvidia GPU. Also, do you have any historic data to compare the Project's performance on OpenCL VS CUDA?

[mgetz]

Can you please tell us what the GPU model is? Based on the log above, it seems to be a Nvidia GPU. Also, do you have any historic data to compare the Project's performance on OpenCL VS CUDA?

NVidia RTX2080 in the log; but I've seen this on a 2070 Super on Linux as well (both are TU104 cores). As for non CUDA/OpenCL comparison I do not, I believe this project started at approximately the same time they rolled out the core update. My concern is just how big a disparity it seems to be against other WUs. As the PPD is literally half that of all other observed projects.

[PantherX]

The difference between PPD for OpenCL VS CUDA can be anywhere between 15% to 100% which is acceptable since any performance increase (if any) by CUDA is considered as a bonus. The benchmark is done on OpenCL to ensure fair points between different devices. There's a 10% variation within that baseline due to different GPUs and their architecture.

My GTX 1080 Ti on Windows (not dedicated) seems to be within the CUDA range:
Min. Time / Frame : 00:00:23 - 2,053,727.82 PPD
Avg. Time / Frame : 00:00:25 - 1,812,277.78 PPD

[mgetz]

WU 16921 (60, 47, 77) seems to replicate the behavior.
Currently running it. So I'd understand that but this is literally less than 1/2 the PPD it's getting with other WUs. On an RTX2080... that makes no sense. I understand there is some points inflation due to COVID, but that wouldn't account for this. I'd expect that to drop down to the 2m PPD range at worst. Particularly since this WU is running in CUDA mode.

I guess what I'm saying here is that there is no way this correlates with the benchmark. Something is seriously wrong with how this WU is running in CUDA mode because it appears to be a net regression.

[NormalDiffusion]

I see the same on an AMD Vega FE and on 2 different R VII: PPD less than 1/2 on what the cards are getting (on one r vii it's less than 1/3...) for other projects.

1080Ti and Titan Xp (under Linux) are showing "normal" PPD.

[jcabana]

The problem appear for all 169XX Working units. It seems that some of theses projects are on the low side as number of atoms goes, and the GPU card are not fully used for these.

elsewhere, Bruce said:

On the other hand small proteins have very few calculations which can be done in parallel and all the serial portions still have to be performed. Your 2060 is great at processing large proteins and not so good working on small proteins.

(about 16904) Is it the case for all 169XX? I do not know.

To which point this apply for 16921, I do not know, but I don't think it's a coincidence that all units from theses projects take longer to do, and make the PPD production drop. It's been like this for a while now. And by the way, I have a RX5700XT. So I doubt that this is a CUDA only problem.

[Neil-B]

You can see the number of atoms in each project on the Project Summary page (linked from the top of the forum) https://apps.foldingathome.org/psummary ... 169xx projects vary in size.

[Kjetil]

https://stats.foldingathome.org/project?p=16921It is cancer(old core 21)

If you want to run covid-19 go to advanced control

[mgetz]

https://stats.foldingathome.org/project?p=16921It is cancer(old core 21)

If you want to run covid-19 go to advanced control

Project summary shows it as core 22, and so does the posted log

https://apps.foldingathome.org/psummary

Also I'd rather not discriminate, while I am favoring donating to COVID causes that doesn't mean I'm opposed to donating to other causes. What I do want to ensure is that things aren't regressing.

You can see the number of atoms in each project on the Project Summary page (linked from the top of the forum) https://apps.foldingathome.org/psummary ... 169xx projects vary in size.

The problem appear for all 169XX Working units. It seems that some of theses projects are on the low side as number of atoms goes, and the GPU card are not fully used for these.

I haven't observed anything else for other 169xx projects, AFAIK this is 16921 only.

[jcabana]

I haven't observed anything else for other 169xx projects, AFAIK this is 16921 only.

For reference, I have taken note of significant PPD drop with 16918, 16905, 16904, 16921 (theses are the one that I have noticed) On a non-CUDA card (AMD). In facts, most of the days my production dropped below 800 000 PPD, I had two of theses unit in one day. My card should be able to handle 1.2 Million PPD per day.

[PantherX]

https://stats.foldingathome.org/project?p=16921It is canser(old core 21)

If you want to run covid-19 go to advansed control

Project summary shows it as core 22, and so does the posted log

https://apps.foldingathome.org/psummary ...

Please note the COVID-19 Projects can be run on FahCore_22 or FahCore_a7 or FahCore_a8.

You will need to look at the individual Projects (https://stats.foldingathome.org/project?p=11747) and in there, you can see the cause set to "covid-19" while others will have different causes or nothing at all which are all valid.

[mgetz]

For reference, I have taken note of significant PPD drop with 16918

I can't speak for AMD but on my 2070 Super on CUDA 16918 (113, 34, 109) is running at 2.56mPPD which is within expected deviance based on observed average variance for CUDA cores on that setup. At least for CUDA cores it seems to be 16921 alone.

[bruce]

1) The problem with projects with low atom counts on a GPU which performs well when a project is capable of being highly parallelized has not changed. WUs tha make good use of a large number of shaders will always see reduced performance when the protein can't make efficient use of all of them. Proteins will always vary and the projects that happen to be assignable at a specific time may be less productive. (See 3 below.)

2) The purpose of the benchmarking project is to make optimum assignments of WUs to GPUs. That project has not yet been implemented yet. Data is still being gathered -- which includes making non-optimum assignments so they know what your GPU can actually produce. That's NOT regressing; it's data gathering

3) Even when the benchmarking project is fully implemented, there will be wide variations. If the only projects available happen to be non-optimum, the two choices are makeing a non-optimum assignment or tell you that there are no projects that fit your hardware. Would you rather have your GPU idle or earning points at a lower rate?

...My card should be able to handle 1.2 Million PPD per day.

4) FAH is guilty of points inflation and it changes your expectations. I have no specific knowledge about your GPU, but maybe its average production should be lower than your expectations I'm sure you only want the highest numbers you've ever seen, even if they happened to be higher than they should be.

[mgetz]

2) The purpose of the benchmarking project is to make optimum assignments of WUs to GPUs. That project has not yet been implemented yet. Data is still being gathered -- which includes making non-optimum assignments so they know what your GPU can actually produce. That's NOT regressing; it's data gathering

I was under the impression based on the FAQ that all projects were run on a standard benchmarking setup and points were determined from that. Is the point allocation method changing?