Folding Forum

toTOW wrote:

Scarlet-Tech wrote:

bruce wrote:At this time, the GPU VRAM clock rate seems to be more important than the Core clock-rate.l

Bcavnaugh is showing that the VRAM has been lowered from stock 7000mhz to 6000mhz and lower.. Well lower than stock speeds and even lower than last generation speeds.

Be careful : most Maxwell2 GPUs run in P2 mode when running FAH, but most overclocking tool (MSI Afterburner, ...) changes only the P0 mode clocks.

You need to use NV Inspector to check which mode is used while running FAH (on my 980M, it's P0, but it's P2 on my 980). Then, you have to change the memory clocks in the right mode.

Some other weird things might occur : on my 980, I have to change memory clocks for P2 tab, but GPU clicks in P0 tab.

No one should have to down clock their GPUs to Fold.

bruce wrote:Stop testing until there's a new version of Core_21 that closes a number of open issues.

When you DECIDE you want Core_21 assignments, add "advanced" As I understand it, all Core_21 projects are now Adv (or even further removed from general circulation). If one slips through, whether it fails or not, let us know what project it is.

I still get Core 21 without the advanced being set as posted above.
When I stated Folding I applied for being a Beta Test as was at that time told no that I did not have the right type of hardware need for testing.
I well complete two more test rounds, a third and forth with the third setting the P2 memory down 1000 MHz and then the forth down 2000 MHz using NvidiaInspector
Then I will stop testing

I guess I'll be getting almost all x18 from the look of it, I'll return my clocks to normal then. but I'll keep an eye out and see if it feeds me any x21, but for now it shouldn't with my current lack of flags.

z999z3mystorys wrote:I guess I'll be getting almost all x18 from the look of it, I'll return my clocks to normal then. but I'll keep an eye out and see if it feeds me any x21, but for now it shouldn't with my current lack of flags.

I am still getting Core 21 UNKNOWN_ENUM P9704 Projects without the Client Type Set.

Have you re-booted your system since changing the client-type?

bcavnaugh wrote: I am still getting Core 21 UNKNOWN_ENUM P9704 Projects without the Client Type Set.

I don't think all Core_21s have been blocked from Maxwell-2 cards without a client-type defined, only the newer ones that have been causing trouble.

Maybe a mod could clarify?

parkut wrote:Have you re-booted your system since changing the client-type?

Yes, and also between Test Rounds.

billford wrote:

bcavnaugh wrote: I am still getting Core 21 UNKNOWN_ENUM P9704 Projects without the Client Type Set.

I don't think all Core_21s have been blocked from Maxwell-2 cards without a client-type defined, only the newer ones that have been causing trouble.

Maybe a mod could clarify?

From Bruce:

bruce wrote:Stop testing until there's a new version of Core_21 that closes a number of open issues.

When you DECIDE you want Core_21 assignments, add "advanced" As I understand it, all Core_21 projects are now Adv (or even further removed from general circulation). If one slips through, whether it fails or not, let us know what project it is.

So it looks like Core 21 UNKNOWN_ENUM P9704 Projects are slipping through.

Ah, right, seems I misunderstood... I'll keep an eye on my rigs.

When I removed the flag from the GPU slots I forgot I'd earlier set advanced under Expert settings for the client so my observations are worth zilch. I didn't really notice as I didn't get any P96xx WUs, but that must have been just a statistical fluke.

9704 runs rather well technically, at least on my Linux-970 ...

Round Three on the GTX 980 HC Rig.
I dropped the Memory Down to 2004 MHz from 3004 MHz (This dropped my PPD By over 580,000)
And I did get a Core 21 UNKNOWN_ENUM Project P9704 My HFM Site http://weather.mfc-cs.com/haf/

According to psummary, the UNKNOWN_ENUM projects are 97xx or 1049x. Which are you getting?
Was it before or after this was posted?

bruce wrote:According to psummary, the UNKNOWN_ENUM projects are 97xx or 1049x. Which are you getting?
Was it before or after this was posted?

This is posted above as well and in the log file posted on Google
And I did get a Core 21 UNKNOWN_ENUM Project P9704
You can find the error project here: https://docs.google.com/document/d/1x6H ... sp=sharing

Last Completed without error (With memory Set to 2004 MHz)
01:49:07:WU01:FS02:0x21:Completed 640000 out of 640000 steps (100%)
01:49:38:WU01:FS02:0x21:Saving result file logfile_01.txt
01:49:38:WU01:FS02:0x21:Saving result file checkpointState.xml
01:49:42:WU01:FS02:0x21:Saving result file checkpt.crc
01:49:42:WU01:FS02:0x21:Saving result file log.txt
01:49:42:WU01:FS02:0x21:Saving result file positions.xtc
01:49:43:WU01:FS02:0x21:Folding@home Core Shutdown: FINISHED_UNIT
01:49:44:WU01:FS02:FahCore returned: FINISHED_UNIT (100 = 0x64)
01:49:44:WU01:FS02:Sending unit results: id:01 state:SEND error:NO_ERROR project:9704 run:59 clone:12 gen:96 core:0x21 unit:0x00000088ab404162553ec4ef9344ea8f
01:49:44:WU01:FS02:Uploading 15.02MiB to 171.64.65.98

It's not when the WU was credited that matters, it's when the WU was assigned.
mpharrigan posted at 4:26 pm (PST)
WU assigned to donor at: 2015-11-20 14:06:05 (PST)
Your WU (P9704 R59 C12 G96) was added to the stats database on 2015-11-20 18:07:59 for 44046 points of credit.

ChristianVirtual wrote:9704 runs rather well technically, at least on my Linux-970 ...

Same here on my 980's... as do the other Core_21's except the 96xx series, but I suppose PG have a lot more Bad State stats to look at than we have!