×
For help and support with ALLBenchmark in general
Question Disappearing results
- Tugrul_512bit
- Offline
- Senior forum
Less
More
- Posts: 77
- Karma: 1
- Thank you received: 16
11 years 7 months ago #492
by Tugrul_512bit
I did 1200/1320 test on kitty, cat and tiger settings. Then i did 1250/1375 only tiger. When i looked at kitty and cat results on allbenchmark site, i saw 1250/1375 on them. Intentionally or a minor flaw ? .
Please Log in or Create an account to join the conversation.
- Tugrul_512bit
- Offline
- Senior forum
Less
More
- Posts: 77
- Karma: 1
- Thank you received: 16
11 years 7 months ago #493
by Tugrul_512bit
Hello,
I just did @ 663/810 on tiger preset and got 3600 points, ok, but the other old results settings are seen as 663/810 at cat and kitty settings. For example I am 9900 points at cat settings with 1200/1320 but the bug changed it to 663/810 just under a user with more advanced gpu @(925/1500). Okay, lucid-virtu users can get %10-%15 more points but i can make myself look like i get similar points with poor frequencies with sacrificing one of the settings(tiger in this example 3600 points to have seat between 7770 users).
I am using 7870 pitcairn.
Thanks.
I just did @ 663/810 on tiger preset and got 3600 points, ok, but the other old results settings are seen as 663/810 at cat and kitty settings. For example I am 9900 points at cat settings with 1200/1320 but the bug changed it to 663/810 just under a user with more advanced gpu @(925/1500). Okay, lucid-virtu users can get %10-%15 more points but i can make myself look like i get similar points with poor frequencies with sacrificing one of the settings(tiger in this example 3600 points to have seat between 7770 users).
I am using 7870 pitcairn.
Thanks.
Please Log in or Create an account to join the conversation.
- Tugrul_512bit
- Offline
- Senior forum
Less
More
- Posts: 77
- Karma: 1
- Thank you received: 16
11 years 7 months ago - 11 years 7 months ago #494
by Tugrul_512bit
Hello,
I'm planning to bench @ 1325/1640 later if this bug will be fixed. Anyway, which catalyst should i use? Which are you working on?
I'm planning to bench @ 1325/1640 later if this bug will be fixed. Anyway, which catalyst should i use? Which are you working on?
Last Edit: 11 years 7 months ago by Tugrul_512bit.
The following user(s) said Thank You: Michal 'bonzaj' Staniszewski
Please Log in or Create an account to join the conversation.
- Michal 'bonzaj' Staniszewski
- Offline
- Moderator
Less
More
- Posts: 263
- Karma: 18
- Thank you received: 61
11 years 7 months ago #499
by Michal 'bonzaj' Staniszewski
Hi!
Yes we are aware of that bug. Unfortunatelly, all people responsible for this are still in California after visiting GDC. They will be back by the end of this week and will take care of it.
Thanks!
Yes we are aware of that bug. Unfortunatelly, all people responsible for this are still in California after visiting GDC. They will be back by the end of this week and will take care of it.
Thanks!
Please Log in or Create an account to join the conversation.
- Tugrul_512bit
- Offline
- Senior forum
Less
More
- Posts: 77
- Karma: 1
- Thank you received: 16
11 years 7 months ago - 11 years 7 months ago #500
by Tugrul_512bit
1625 MHz vram is stable in unigine valley 1.0.
1550MHz flickers here.
1500 MHz is stable in here.
This benchmark really checks vram stability. There's a part where two beams crash each other, makes gpu 4-5 °C hotter instantly (kind of furmark) which makes the part a good thermal load tester.
Card's vram is Hynix-t2c of version 1250 MHz @1.5V but hawk comes with 1200MHz @1.6V.
Should I increase voltage to test for 1600 MHz? Because its already increased by 1V. Dangerous?
I'm asking 1600MHz because there is a %10 performance at 1376 MHz compared to 1375MHz. Its kind of Cas Latency loosening which i cannot control.
1550MHz flickers here.
1500 MHz is stable in here.
This benchmark really checks vram stability. There's a part where two beams crash each other, makes gpu 4-5 °C hotter instantly (kind of furmark) which makes the part a good thermal load tester.
Card's vram is Hynix-t2c of version 1250 MHz @1.5V but hawk comes with 1200MHz @1.6V.
Should I increase voltage to test for 1600 MHz? Because its already increased by 1V. Dangerous?
I'm asking 1600MHz because there is a %10 performance at 1376 MHz compared to 1375MHz. Its kind of Cas Latency loosening which i cannot control.
Last Edit: 11 years 7 months ago by Tugrul_512bit.
The following user(s) said Thank You: Michal 'bonzaj' Staniszewski
Please Log in or Create an account to join the conversation.
- Michal 'bonzaj' Staniszewski
- Offline
- Moderator
Less
More
- Posts: 263
- Karma: 18
- Thank you received: 61
11 years 7 months ago #504
by Michal 'bonzaj' Staniszewski
Hi!
Thanks for that! You are completely right here the GPU test (raymarching) is an extreme heater. It's because the GPU utilisation reaches 95% on some configs and we do mostly ALU here. So not exactly typical VRAM test, but the driver has not too much work here. I'll ask around if someone could send you an advanced version. In catzilla mode you may fry your GPU .
Thanks for that! You are completely right here the GPU test (raymarching) is an extreme heater. It's because the GPU utilisation reaches 95% on some configs and we do mostly ALU here. So not exactly typical VRAM test, but the driver has not too much work here. I'll ask around if someone could send you an advanced version. In catzilla mode you may fry your GPU .
Please Log in or Create an account to join the conversation.