Author Topic: Noise lines on screen and crash (Read 6047 times)

Locutus73 · « **Reply #20 on:** April 12, 2017, 12:03:23 pm »

Quote from: ker on April 12, 2017, 11:35:58 am

I bought mine in a Spain local store. It's a GPD Win rev2 Z8750 (2017-03-11 is printed in the box).

Same as mine.

Quote from: ker on April 12, 2017, 11:35:58 am

I received it yesterday so I've not done many tests, but I've not seen a single crash or issue with graphic driver.

Bios: 2.17.1249
Intel(R) HD Graphics drivers: 20.19.15.4331

This seems the same BIOS as the one used by MikeD who isn't reporting issues for now.
To be honest I had an immediate crash upon the first start with the original installation.
Since I don?t trust Chinese Windows installations with preconfigured users, I immediately erased the disk (with gparted live) and proceeded with a clean install using 20.19.15.5439 beta drivers.

Quote from: ker on April 12, 2017, 11:35:58 am

Whan I enter bios I don't see any options at all (just basic ones) and screen is rotated.

The rotation issue on BIOS is normal.
You have just basic settings because you?re on 2016.11.18 BIOS; you must flash (at your risk) 2016.10.25 BIOS in order to get advanced settings.
I experienced issues both on 2016.11.18 BIOS (with no memory settings available, so memory at 1600MHz) and on 2016.10.25 BIOS (with memory set at 1600MHz).

Quote from: ker on April 12, 2017, 11:35:58 am

I can do some tests if you need to...

I get quite repeatable crashes using free 3DMark benchmark, doing the ICE STORM (not the EXTREME one) test. After the initial demo, I often experience crashes at the first benchmark. See https://youtu.be/03GwBdfB5r8
Sometimes my GPD WIN goes through the test flawlessly many times in a row, but if I insist, I get the crash. Usually I get the crash if I run the benchmark with Windows just booted.

Many thanks in advance.
Best regards.

Locutus73

Locutus73 · « **Reply #21 on:** April 12, 2017, 12:06:39 pm »

Quote from: Maniac on April 12, 2017, 11:58:09 am

Can you guys run the Windows 10 memory test with your ram set to 1600?
http://www.thewindowsclub.com/windows-memory-diagnostics-tool-in-windows-7

Done in the past (even with DRAM set to fast if I recall correctly) and passed with no problems.

Best regards.
Locutus73

Maniac · « **Reply #22 on:** April 12, 2017, 12:16:07 pm »

Quote from: Locutus73 on April 12, 2017, 12:06:39 pm

Quote from: Maniac on April 12, 2017, 11:58:09 am
Can you guys run the Windows 10 memory test with your ram set to 1600?
http://www.thewindowsclub.com/windows-memory-diagnostics-tool-in-windows-7
Done in the past (even with DRAM set to fast if I recall correctly) and passed with no problems.

Best regards.
Locutus73

What about memtest 86?
http://www.memtest86.com/download.htm

ker · « **Reply #23 on:** April 12, 2017, 01:00:02 pm »

Quote from: Locutus73 on April 12, 2017, 12:03:23 pm

I get quite repeatable crashes using free 3DMark benchmark, doing the ICE STORM (not the EXTREME one) test. After the initial demo, I often experience crashes at the first benchmark. See https://youtu.be/03GwBdfB5r8
Sometimes my GPD WIN goes through the test flawlessly many times in a row, but if I insist, I get the crash. Usually I get the crash if I run the benchmark with Windows just booted.

Many thanks in advance.
Best regards.

Locutus73

I've tried this 3DMark ICE STORM test 3 times in a row. Not issues.

First row:
Total score: 24209
Graphics score: 25927 (114.29 and 111.21 FPS)
Physics score: 19652 (62.39 FPS)

Second row:
Total score: 24144
Graphics score: 25751 (111.82 and 112.11 FPS)
Physics score: 19816 (62.91 FPS)

Third row:
Total score: 23628
Graphics score: 25089 (110.18 and 108.01 FPS)
Physics score: 19628 (62.31 FPS)

Locutus73 · « **Reply #24 on:** April 12, 2017, 02:08:39 pm »

Quote from: ker on April 12, 2017, 01:00:02 pm

I've tried this 3DMark ICE STORM test 3 times in a row. Not issues.

First row:
Total score: 24209
Graphics score: 25927 (114.29 and 111.21 FPS)
Physics score: 19652 (62.39 FPS)

Second row:
Total score: 24144
Graphics score: 25751 (111.82 and 112.11 FPS)
Physics score: 19816 (62.91 FPS)

Third row:
Total score: 23628
Graphics score: 25089 (110.18 and 108.01 FPS)
Physics score: 19628 (62.31 FPS)

Did you run Icestorm Unlimited (the white background button with orange text) in order to get these results? With Icestorm Unlimeted I get results in this order of magnitude (approximately 26000) and no crashes, but with the regular Icestorm (the orange button with white text) I get approximately 15000 and random crashes.

Many thanks in advance.
Best regards.

Locutus73

ker · « **Reply #25 on:** April 12, 2017, 04:05:52 pm »

Quote from: Locutus73 on April 12, 2017, 02:08:39 pm

Did you run Icestorm Unlimited (the white background button with orange text) in order to get these results? With Icestorm Unlimeted I get results in this order of magnitude (approximately 26000) and no crashes, but with the regular Icestorm (the orange button with white text) I get approximately 15000 and random crashes.

Many thanks in advance.
Best regards.

Locutus73

White one. I've just tested the orange one twice. No crashes and no difference... about the same results: 24000 aprox

Locutus73 · « **Reply #26 on:** April 12, 2017, 04:59:41 pm »

Quote from: ker on April 12, 2017, 04:05:52 pm

White one. I've just tested the orange one twice. No crashes and no difference... about the same results: 24000 aprox

Uhm... 24000???
Do you see differences between the two tests?
I mean: they?re completely different
1) IceStorm Unlimited: during tests you see a central window, on a static background, with strange animations going on; no real 3D animations on screen. See https://www.youtube.com/watch?v=Cl4fyftyY1Y
2) Regular IceStorm: during test you have real 3D animations like in a space video game. See https://www.youtube.com/watch?v=ZQh5vv998uk

Many thanks in advance,
best regards.

Locutus73

Locutus73 · « **Reply #27 on:** April 12, 2017, 08:00:09 pm »

Quote from: Locutus73 on April 12, 2017, 04:59:41 pm

Quote from: ker on April 12, 2017, 04:05:52 pm
White one. I've just tested the orange one twice. No crashes and no difference... about the same results: 24000 aprox

Uhm... 24000???
Do you see differences between the two tests?
I mean: they?re completely different
1) IceStorm Unlimited: during tests you see a central window, on a static background, with strange animations going on; no real 3D animations on screen. See https://www.youtube.com/watch?v=Cl4fyftyY1Y
2) Regular IceStorm: during test you have real 3D animations like in a space video game. See https://www.youtube.com/watch?v=ZQh5vv998uk

Many thanks in advance,
best regards.

Locutus73

I got 26000 on regular IceStorm too performing the benchmark connected to the AC. I previously got 15000 under battery.
I even completed a complete benchmark run with memory set at 1600MHz-Fast with no crashes for the first time (under AC).

I have to investigate further.

Locutus73

Locutus73 · « **Reply #28 on:** April 13, 2017, 05:42:00 pm »

Quote from: Maniac on April 12, 2017, 12:16:07 pm

What about memtest 86?
http://www.memtest86.com/download.htm

Done and passed in around 5 hours, with memory at 1600Mhz Fast.

My findings so far:
1)   With memory set at 1067 Fast the system seems to be rock solid
2)   With memory set at 1600 Fast
      a.   On alternate current 3DMark IceStorm seems to be stable
      b.   On battery with Intel Driver->Graphics Properties->Power->Extended Battery Life for Gaming ENABLED, 3DMark IceStorm frequently crashes as in https://youtu.be/03GwBdfB5r8
      c.   On battery with Intel Driver->Graphics Properties->Power->Extended Battery Life for Gaming DISABLED, 3DMark IceStorm seems to be stable (and I get higher results, around 27000 instead of 15000)
      d.   Prime95 seems stable
      e.   Geekbench 4 randomly produces vertical artifacts in Desktop as in https://youtu.be/03GwBdfB5r8 and sometime crashes into a bluescreen
3)   With memory set at 1600 Typical: work in progress?

Any ideas regarding the correlation between memory speed 1600 Fast and "Extended Battery Life for Gaming" ENABLED (so lower performances) on battery causing 3D rendering crashes?

Best regards
Locutus73

Maniac · « **Reply #29 on:** April 14, 2017, 12:25:23 pm »

Oh, it's been reported numerous times that "Extended Battery Life for Gaming" can cause issues. Leave that disabled.

Locutus73 · « **Reply #30 on:** April 14, 2017, 01:13:11 pm »

Quote from: Maniac on April 14, 2017, 12:25:23 pm

Oh, it's been reported numerous times that "Extended Battery Life for Gaming" can cause issues. Leave that disabled.

You are right (I believed that it was just a performance hint, not a stability fix); but this parameter doesn?t resolve the Geekbench crashes. The system is still unstable
BUT
I have a great update!
Continuing with my investigation I found that using the following BIOS setting
Chipset Tab -> North Bridge -> Memory Configuration -> DRAM PM5 -> Disabled
the system seems to be rock solid even with memory set at 1600MHz Fast.

So, with Memory set at 1600MHz Fast and PM5 disabled I have
1) 3DMark IceStorm stable regardless of power source (battery and AC) and Intel Driver ?Extended Battery Life for Gaming? mode
2) Geekbench 4 stable
I even tried to run 3DMark and Geekbench at the same time with no problems.

So it seems that disabling DRAM PM5 leads to a perfect stable system regardless of Windows/drivers settings.

This DRAM PM5 parameter is poorly documented (the BIOS states ?Enable or disable DRAM PM5 PUNIT configuration.?), but digging around PM5 seems to be an advanced power saving mode for DRAM. My hypothesis is that, disabling this parameter, the system uses higher voltages for the DDR3LP memories (or, for the sake of precision, the system avoids lower voltages for the memories) avoiding possible instability at 1600MHz. After all, when we overclock CPUs, GPUs and RAMs, we obtain stability using higher voltages. The hypothesis is compatible with the observed correlation between crashes and Intel "Extended Battery Life for Gaming" parameter: it is possible that, in this mode, all power saving technologies kick in frequently, leading to lower voltage and instability.

Now, can bitersnake, MikeD, nitty and Blm79 verify if, using memory set at 1600MHz Fast and PM5 disabled, the vertical artifacts and bluescreen issue is fixed?

It would be interesting to determine if disabling ?DRAM PM5? resolves the driver freezes (not the total crashes like mine) and, in general, makes the GPD WIN more (or totally) stable.

Best regards
Locutus73

Maniac · « **Reply #31 on:** April 14, 2017, 03:29:27 pm »

Nice find!

nitty · « **Reply #32 on:** April 14, 2017, 06:06:59 pm »

Quote from: Locutus73 on April 14, 2017, 01:13:11 pm

Quote from: Maniac on April 14, 2017, 12:25:23 pm
Oh, it's been reported numerous times that "Extended Battery Life for Gaming" can cause issues. Leave that disabled.

You are right (I believed that it was just a performance hint, not a stability fix); but this parameter doesn?t resolve the Geekbench crashes. The system is still unstable
BUT
I have a great update!
Continuing with my investigation I found that using the following BIOS setting
Chipset Tab -> North Bridge -> Memory Configuration -> DRAM PM5 -> Disabled
the system seems to be rock solid even with memory set at 1600MHz Fast.

So, with Memory set at 1600MHz Fast and PM5 disabled I have
1) 3DMark IceStorm stable regardless of power source (battery and AC) and Intel Driver ?Extended Battery Life for Gaming? mode
2) Geekbench 4 stable
I even tried to run 3DMark and Geekbench at the same time with no problems.

So it seems that disabling DRAM PM5 leads to a perfect stable system regardless of Windows/drivers settings.

This DRAM PM5 parameter is poorly documented (the BIOS states ?Enable or disable DRAM PM5 PUNIT configuration.?), but digging around PM5 seems to be an advanced power saving mode for DRAM. My hypothesis is that, disabling this parameter, the system uses higher voltages for the DDR3LP memories (or, for the sake of precision, the system avoids lower voltages for the memories) avoiding possible instability at 1600MHz. After all, when we overclock CPUs, GPUs and RAMs, we obtain stability using higher voltages. The hypothesis is compatible with the observed correlation between crashes and Intel "Extended Battery Life for Gaming" parameter: it is possible that, in this mode, all power saving technologies kick in frequently, leading to lower voltage and instability.

Now, can bitersnake, MikeD, nitty and Blm79 verify if, using memory set at 1600MHz Fast and PM5 disabled, the vertical artifacts and bluescreen issue is fixed?

It would be interesting to determine if disabling ?DRAM PM5? resolves the driver freezes (not the total crashes like mine) and, in general, makes the GPD WIN more (or totally) stable.

Best regards
Locutus73

Can confirm that these changes work! My system is stable after turning the RAM speed back to 1600 with DRAM PM5 disabled. While I haven't been able to test it too much, it also seems to have fixed random crashes after waking from sleep(these occurred even with the RAM speed set at 1067) . Will need to test it some more over the next couple days to be sure though.

Thanks!

bitersnake · « **Reply #33 on:** April 14, 2017, 08:15:39 pm »

Quote from: Locutus73 on April 14, 2017, 01:13:11 pm

Quote from: Maniac on April 14, 2017, 12:25:23 pm
Oh, it's been reported numerous times that "Extended Battery Life for Gaming" can cause issues. Leave that disabled.

You are right (I believed that it was just a performance hint, not a stability fix); but this parameter doesn?t resolve the Geekbench crashes. The system is still unstable
BUT
I have a great update!
Continuing with my investigation I found that using the following BIOS setting
Chipset Tab -> North Bridge -> Memory Configuration -> DRAM PM5 -> Disabled
the system seems to be rock solid even with memory set at 1600MHz Fast.

So, with Memory set at 1600MHz Fast and PM5 disabled I have
1) 3DMark IceStorm stable regardless of power source (battery and AC) and Intel Driver ?Extended Battery Life for Gaming? mode
2) Geekbench 4 stable
I even tried to run 3DMark and Geekbench at the same time with no problems.

So it seems that disabling DRAM PM5 leads to a perfect stable system regardless of Windows/drivers settings.

This DRAM PM5 parameter is poorly documented (the BIOS states ?Enable or disable DRAM PM5 PUNIT configuration.?), but digging around PM5 seems to be an advanced power saving mode for DRAM. My hypothesis is that, disabling this parameter, the system uses higher voltages for the DDR3LP memories (or, for the sake of precision, the system avoids lower voltages for the memories) avoiding possible instability at 1600MHz. After all, when we overclock CPUs, GPUs and RAMs, we obtain stability using higher voltages. The hypothesis is compatible with the observed correlation between crashes and Intel "Extended Battery Life for Gaming" parameter: it is possible that, in this mode, all power saving technologies kick in frequently, leading to lower voltage and instability.

Now, can bitersnake, MikeD, nitty and Blm79 verify if, using memory set at 1600MHz Fast and PM5 disabled, the vertical artifacts and bluescreen issue is fixed?

It would be interesting to determine if disabling ?DRAM PM5? resolves the driver freezes (not the total crashes like mine) and, in general, makes the GPD WIN more (or totally) stable.

Best regards
Locutus73

Tbh I hadn't experienced the screen distortion since over a week and while I have made the change to the setting you have recommended here, my results cannot be treated as reliable-- I never had stability issues on my device so everyone else' s case here seems different from mine. Nevertheless, this quite informative and something that I recommend you also post on the GPD win reddit (once folks here can verify that this does actually solve the issue).

MJPIA · « **Reply #34 on:** April 14, 2017, 11:18:01 pm »

Quote from: Locutus73 on April 12, 2017, 11:18:33 am

Quote from: MJPIA on April 11, 2017, 10:50:02 pm
A bunch of people that received theirs around the same time period with the same issues?
One or two people could be a coincidence.
5 people though?
I wonder if we've got a bad batch here.
5 people are not sufficient for scientific stats, but we are free to make guesses.
It seems that the March GearBest GPD WIN rev2 Z8750 batch is affected by low grade memories that are not able to run at 1600MHz; we can even suppose that all the widespread bugs seen in the past are memory related. If the hypothesis is right, we should concentrate on memory.
How much does 1067MHz DRAM affects performance vs 1600MHz DRAM?
Does the ?resistor fix? increase DRAM voltage leading to a more stable system?

Many thanks in advance.
Best regards.

Locutus73

I know I'm late responding and you've already found some potential fixes but the resistor fix was for CPU voltage and in all the back and forth I had with GPD nothing was ever mentioned about RAM voltage.
I know changing ram speed did absolutely nothing for the graphics crashes.

Anyways yeah PM5 does seem to be a RAM power saving mode of some sort.
https://patchwork.kernel.org/patch/5807271/
Different OS but Intel related BIOS

Quote

From: Ville Syrj?l? <[email protected]>

CHV has a new knob in Punit to select between some memory power savings
modes PM2 and PM5. We can allow the deeper PM5 when maxfifo mode is
enabled, so let's do so in the hopes for moar power savings.

Given PM5 is a deeper power saving mode if a batch of RAM has some issues causing problems during periods of inactivity is one thing but I'm not sure why this would affect people in the middle of intensive benchmarks.
Given the lack of documentation of Punit functions though who knows?
Hopefully this is a longterm permanent fix for those with problems.

Maniac · « **Reply #35 on:** April 14, 2017, 11:43:40 pm »

Quote from: nitty on April 14, 2017, 06:06:59 pm

Can confirm that these changes work! My system is stable after turning the RAM speed back to 1600 with DRAM PM5 disabled. While I haven't been able to test it too much, it also seems to have fixed random crashes after waking from sleep(these occurred even with the RAM speed set at 1067) . Will need to test it some more over the next couple days to be sure though.

Thanks!

I was wondering if it would fix that! Now to decide if I want to re-enable sleep (which is a bit of pain because connected standby ignores wake timers) or just stick with hibernation.

Blm79 · « **Reply #36 on:** April 15, 2017, 11:44:39 am »

Hello guys, sorry for being late, i was busy this week.
I discovered something interesting: the artifacts comes only when in mouse mode! Here is a short video i made demonstrating this: https://www.youtube.com/watch?v=lW-kKaLTAt8

However, I have not found the PM5 option in the bios. Should I flash the other BIOS?
EDIT: Where i find older BIOS and instructions ?

Maniac · « **Reply #37 on:** April 15, 2017, 03:38:41 pm »

Yes you need the 10/25 bios.
https://www.reddit.com/r/gpdwin/comments/5c9duk/gpd_win_general_faq/

Blm79 · « **Reply #38 on:** April 15, 2017, 05:00:29 pm »

Thank's @Maniac, i downgraded the bios successfully.
@Locutus73 I set the options like you said (using memory set at 1600MHz Fast and PM5 disabled) and i can confirm that it works flawlessly !
On the other hand I don't understand the link with the mouse...

Maniac · « **Reply #39 on:** April 17, 2017, 01:16:54 pm »

Quote from: Maniac on April 14, 2017, 11:43:40 pm

Quote from: nitty on April 14, 2017, 06:06:59 pm
Can confirm that these changes work! My system is stable after turning the RAM speed back to 1600 with DRAM PM5 disabled. While I haven't been able to test it too much, it also seems to have fixed random crashes after waking from sleep(these occurred even with the RAM speed set at 1067) . Will need to test it some more over the next couple days to be sure though.

Thanks!

I was wondering if it would fix that! Now to decide if I want to re-enable sleep (which is a bit of pain because connected standby ignores wake timers) or just stick with hibernation.

I tried this out. I still got crashes during sleep, so this isn't my issue.

Author Topic: Noise lines on screen and crash (Read 6047 times)

bitersnake (OP)