Topic Review (Newest First) |
10-29-2019 06:10 PM | |
skupples |
![]() I'll miss these 1080tis, but considering I half killed both, good riddance. I ALMOST went with a $300 used 1080ti instead, but meh. Seems Pascal just wasn't for me / and or these cards are too fragile for used action. 5700xt til 4080ti, or bust. |
10-29-2019 06:08 PM | |
neurotix |
Benchmarking trick.
Install this. https://github.com/abbodi1406/vcredi...ses/tag/0.26.0 Run Nvidia Profile Inspector as Admin I added here as a .zip. Put this to Off, hit Apply Changes TWICE! ![]() Set your RAM speed in Afterburner. Run the runCustom.bat included in the second .zip file folder with the .exe etc. left in the folder. Raise your video card RAM speed one notch at a time, rerun the bandwidth tester. You'll find on some memory settings you RAM will measure say 420 to 430 Gbyte/sec but a notch or two higher or lower it'll be at 470-490 Gbyte/sec. Run your video card benchmarks when it reads the higher bandwidth, your benchmark will do better, even if your actual memory clocks are one or two notches lower. Will definitely be trying this and something else I have up my sleeve for my Ryzen setup. Rep+ thanks, will be seeing if this makes a difference or not. |
10-27-2019 11:03 PM | |
KedarWolf |
Benchmarking trick. Install this. https://github.com/abbodi1406/vcredi...ses/tag/0.26.0 Run Nvidia Profile Inspector as Admin I added here as a .zip. Put this to Off, hit Apply Changes TWICE! ![]() Set your RAM speed in Afterburner. Run the runCustom.bat included in the second .zip file folder with the .exe etc. left in the folder. Raise your video card RAM speed one notch at a time, rerun the bandwidth tester. You'll find on some memory settings you RAM will measure say 420 to 430 Gbyte/sec but a notch or two higher or lower it'll be at 470-490 Gbyte/sec. Run your video card benchmarks when it reads the higher bandwidth, your benchmark will do better, even if your actual memory clocks are one or two notches lower. |
10-27-2019 07:47 PM | |
ThrashZone |
Hi, Now you're just showing off ![]() |
10-27-2019 05:20 PM | |
KedarWolf |
I don't have Fire Strike on hand; can't find the screenshot. I'll rerun it tomorrow. It was just a hair over 36k, something like 36081. Cinebench was 7700+, I'll post the screenshot later.
Thats a great Heaven score. Submit it if you didn't. You are also within 10 fps of the best 2080ti SLI submission. My cards core clocks won't do much more than that, even with added voltage, which adds heat. The memory on either doesn't go much higher than +425 and going higher reduces fps. Also, I do all benches with both cards fans at 100%. The top one will pass 70c in Heaven, bottom will hang around 55c. Unless I open a window, its already near freezing this time of year where I live. The heat is ok though because in games it is rare for either card to exceed 60c even with a fan curve, I use vsync and frame rate limiting. If I could run all cores of my chip at 5ghz, I'd be much closer, to your scores, but I knew that going in. I'd be interested to see FS Ultra as that is far less cpu dependent. Mine was a little more than 14k. https://www.3dmark.com/fs/20798900 FireStrike Ultra 14555 https://www.3dmark.com/fs/20798884 |
10-27-2019 04:58 PM | |
KedarWolf |
I don't have Fire Strike on hand; can't find the screenshot. I'll rerun it tomorrow. It was just a hair over 36k, something like 36081. Cinebench was 7700+, I'll post the screenshot later.
Thats a great Heaven score. Submit it if you didn't. You are also within 10 fps of the best 2080ti SLI submission. My cards core clocks won't do much more than that, even with added voltage, which adds heat. The memory on either doesn't go much higher than +425 and going higher reduces fps. Also, I do all benches with both cards fans at 100%. The top one will pass 70c in Heaven, bottom will hang around 55c. Unless I open a window, its already near freezing this time of year where I live. The heat is ok though because in games it is rare for either card to exceed 60c even with a fan curve, I use vsync and frame rate limiting. If I could run all cores of my chip at 5ghz, I'd be much closer, to your scores, but I knew that going in. I'd be interested to see FS Ultra as that is far less cpu dependent. Mine was a little more than 14k. KedarWolf - i9 9900k/5100MHz - 2x GTX 1080 Ti 2037/6210MHz - 297.5- 7494 - Nvidia Studio drivers. ![]() ![]() |
10-27-2019 10:37 AM | |
ThrashZone |
Hi, Yep nice clock on a very picky heaven benchmark KW ![]() |
10-27-2019 09:49 AM | |
neurotix |
Best run. Try overclocking your memory and use custom curves in Afterburner, CTRL F, max out your CPU clocks and memory at 1093v.
KedarWolf - i9 9900k/5000MHz - 2x GTX 1080 Ti 2037/6237MHz - 294.1- 7407 ![]() ![]() I don't have Fire Strike on hand; can't find the screenshot. I'll rerun it tomorrow. It was just a hair over 36k, something like 36081. Cinebench was 7700+, I'll post the screenshot later. Thats a great Heaven score. Submit it if you didn't. You are also within 10 fps of the best 2080ti SLI submission. My cards core clocks won't do much more than that, even with added voltage, which adds heat. The memory on either doesn't go much higher than +425 and going higher reduces fps. Also, I do all benches with both cards fans at 100%. The top one will pass 70c in Heaven, bottom will hang around 55c. Unless I open a window, its already near freezing this time of year where I live. The heat is ok though because in games it is rare for either card to exceed 60c even with a fan curve, I use vsync and frame rate limiting. If I could run all cores of my chip at 5ghz, I'd be much closer, to your scores, but I knew that going in. I'd be interested to see FS Ultra as that is far less cpu dependent. Mine was a little more than 14k. |
10-27-2019 01:08 AM | |
KedarWolf |
Hey all,
Attachment 302212 I recently got 8th place in the Heaven top 30 thread (at 274 fps), with my CPU @ 4475MHz on the first Core Complex, memory at 3800MHz/1900 fclk and my 1080tis @ 2025/5899 with Kboost on in Precision XOC, stock voltage, 100% fan and side panel/fan filters off. I'm trying to beat that 279 fps and get it as high as possible to beat some people at Heaven My cards are 6696 models and my BIOS is 86.02.39.01.90 (from around March 2018), with no power limit issues (as I can see 127% power usage on both cards in say, looped Fire Strike Ultra combined test). Both are on the slave BIOS. These cards have always overclocked like crap on the core and I don't think it's heat related, if I try and raise both to 2037MHz and max out the voltage, most benches crash to black screen before they can even heat up, and I have to hard reset my rig (hold the power button). In this situation, should I try to flash the EVGA K|NGP|IN bios and possibly shunt mod? Or since the pair is crashing instantly on raising the core frequency (at like 40c top/35c bottom) within a second or two, will this not make a difference because it's not power related? Should I try flashing the BIOS from this thread? Should I also regress drivers to 390.xx if I can on Win10 Pro 1903? (These were the best performing drivers before RTX dropped) I want to redo every bench I did on HWBOT with my 1080tis, multi-gpu and single, now that I get a huge boost to overall score because my physics score is like 35k with this new CPU + memory, but want to get my core OC higher if possible Any advice appreciated. Thanks KedarWolf - i9 9900k/5000MHz - 2x GTX 1080 Ti 2037/6237MHz - 294.1- 7407 ![]() ![]() |
10-26-2019 03:41 PM | |
KedarWolf |
Wanna do Cinebench r20 or 3dmark Fire Strike?
lol jk
Spoiler!
For that run, my chip was at 4475/4450/4200/4200MHz, memory was 3800 c16, and the cards were barely at 2012MHz/5940MHz... what was your 9900k and GPUs at? Sig clocks? Anyway... also thanks to the other person who explained the XOC bios and 65C thermal shutoff, glad I came here and asked first or they'd basically be unusable because of that. Glad I left them alone This Cinebench run was at my 24/7 overclock settings. https://www.overclock.net/forum/6-in...l#post28046906 https://www.3dmark.com/3dm/40489500 36873 Fire strike. |
This thread has more than 10 replies. Click here to review the whole thread. |
Posting Rules | |