EVGA

Highest timespy score I can get with 3080 XC3 Ultra

Page: < 12 Showing page 2 of 2
Author
sethleigh
SSC Member
  • Total Posts : 796
  • Reward points : 0
  • Joined: 2015/08/12 11:27:56
  • Status: offline
  • Ribbons : 4
Re: Highest timespy score I can get with 3080 XC3 Ultra 2021/12/26 12:09:47 (permalink)
CraptacularOne2 things, this thread is from December of last year and second, this thread was about RTX 3080s, not 3090s.

Two things: the guy you replied to was the 4th one to reply to this necro-rezzed thread, not the one who rezzed it. Second, since a previous poster had suggested adding a 3090 to the OP's machine in response to OP's possibly rhetorical question asking for tips, there is an articulable nexus to the contents of the thread to the 3090.
 
Since the 3080 series is still current generation, I'm not sure necro-rezzing this thread was necessarily a bad thing. I haven't got a 3080 score to add to this, but if anyone is interested in comparing the 3080 scores with a 3080ti score, here's my best, taken right after I'd installed the Optimus water block and was taking the card out for a spin. It was good for #11 on the day I ran it of scores achieved with a Ryzen 5900x cpu and 3080ti gpu.

Happy EVGA customer.  Affiliate Code: 0Y7-1VU-ATW2
 
GigaByte X570 Aorus Master, AMD Ryzen 5900x under Optimus Foundation block, 32gb G.Skill DDR4 @ 3800 MHz 14-14-14-28, EVGA 3080ti FTW3 Ultra under Optimus block, 2TB 980 Pro SSD, EVGA Supernova G6 850W PS, ASUS 34" 3440x1440p 120Hz ultrawide, Lenovo 24" 1080p secondary monitor, Win 10

#31
Nereus
Captain Goodvibes
  • Total Posts : 18916
  • Reward points : 0
  • Joined: 2009/04/09 20:05:53
  • Location: Brooklyn, NYC.
  • Status: offline
  • Ribbons : 58
Re: Highest timespy score I can get with 3080 XC3 Ultra 2021/12/26 16:51:53 (permalink)
sethleigh
CraptacularOne2 things, this thread is from December of last year and second, this thread was about RTX 3080s, not 3090s.

Two things: the guy you replied to was the 4th one to reply to this necro-rezzed thread, not the one who rezzed it. Second, since a previous poster had suggested adding a 3090 to the OP's machine in response to OP's possibly rhetorical question asking for tips, there is an articulable nexus to the contents of the thread to the 3090.

Since the 3080 series is still current generation, I'm not sure necro-rezzing this thread was necessarily a bad thing. I haven't got a 3080 score to add to this, but if anyone is interested in comparing the 3080 scores with a 3080ti score, here's my best, taken right after I'd installed the Optimus water block and was taking the card out for a spin. It was good for #11 on the day I ran it of scores achieved with a Ryzen 5900x cpu and 3080ti gpu.

Damn, that's impressive with a 3080Ti, your CPU also helped with that great overall score. I have not cracked 20k with my 3090. 19,883.
 


  BUILD 1 2   |   MINI-ITX BUILD   |   MODSRIGS $1K WIN   |   HEATWARE 111-0-0   |   ASSOCIATE CODE CSKKXUT5Q9GVAFR

#32
sethleigh
SSC Member
  • Total Posts : 796
  • Reward points : 0
  • Joined: 2015/08/12 11:27:56
  • Status: offline
  • Ribbons : 4
Re: Highest timespy score I can get with 3080 XC3 Ultra 2021/12/27 08:56:56 (permalink)
Nereus
 
Damn, that's impressive with a 3080Ti, your CPU also helped with that great overall score. I have not cracked 20k with my 3090. 19,883.

Thanks! Yeah, I compared that score to a few others in the same vicinity on the ranking and most of them had slightly better GPU scores than mine, and it was my CPU score that helped carry it quite that high. My 5900x is pushed fairly aggressively using AMD's built-in OCing tools (PBO2/Curve Optimizer), and my RAM profile is all hand-tuned timings that are about as fast as 3800MT/s RAM can go.
 
I'm going to guess that your 3090 is air cooled from the fact that your GPU hit a max speed of 2175MHz (faster than my max 2160MHz for that run) but only maintained an average of 2003MHz vs. my average of 2105MHz. Your 3090 is probably crossing the thermal throttling temperature threshold for parts of the benchmark and slowing down, whereas my water-cooled 3080ti never crosses that threshold so whatever GPU speed it attains in response to the particular load demanded by the benchmark it's able to maintain. It makes a difference in benchmark scores if you care about it, but it's going to be a hard sell arguing that you actually need better during games. If this matters: my system with the 3080ti is vastly quieter than it was during the couple months I ran the 3080ti on air while waiting for Optimus to ship me the water block. The cooling load for the GPU that used those three dinky fans on the 3080ti turning at thousands of RPM was transferred to the three 140mm and two 180mm fans on my radiators turning probably not even 1000 RPM, and that makes a colossal difference.

Happy EVGA customer.  Affiliate Code: 0Y7-1VU-ATW2
 
GigaByte X570 Aorus Master, AMD Ryzen 5900x under Optimus Foundation block, 32gb G.Skill DDR4 @ 3800 MHz 14-14-14-28, EVGA 3080ti FTW3 Ultra under Optimus block, 2TB 980 Pro SSD, EVGA Supernova G6 850W PS, ASUS 34" 3440x1440p 120Hz ultrawide, Lenovo 24" 1080p secondary monitor, Win 10

#33
Nereus
Captain Goodvibes
  • Total Posts : 18916
  • Reward points : 0
  • Joined: 2009/04/09 20:05:53
  • Location: Brooklyn, NYC.
  • Status: offline
  • Ribbons : 58
Re: Highest timespy score I can get with 3080 XC3 Ultra 2021/12/28 19:05:05 (permalink)
sethleigh
Nereus
 
Damn, that's impressive with a 3080Ti, your CPU also helped with that great overall score. I have not cracked 20k with my 3090. 19,883.

Thanks! Yeah, I compared that score to a few others in the same vicinity on the ranking and most of them had slightly better GPU scores than mine, and it was my CPU score that helped carry it quite that high. My 5900x is pushed fairly aggressively using AMD's built-in OCing tools (PBO2/Curve Optimizer), and my RAM profile is all hand-tuned timings that are about as fast as 3800MT/s RAM can go.
 
I'm going to guess that your 3090 is air cooled from the fact that your GPU hit a max speed of 2175MHz (faster than my max 2160MHz for that run) but only maintained an average of 2003MHz vs. my average of 2105MHz. Your 3090 is probably crossing the thermal throttling temperature threshold for parts of the benchmark and slowing down, whereas my water-cooled 3080ti never crosses that threshold so whatever GPU speed it attains in response to the particular load demanded by the benchmark it's able to maintain. It makes a difference in benchmark scores if you care about it, but it's going to be a hard sell arguing that you actually need better during games. If this matters: my system with the 3080ti is vastly quieter than it was during the couple months I ran the 3080ti on air while waiting for Optimus to ship me the water block. The cooling load for the GPU that used those three dinky fans on the 3080ti turning at thousands of RPM was transferred to the three 140mm and two 180mm fans on my radiators turning probably not even 1000 RPM, and that makes a colossal difference.


Nope.. water cooled; full CPU/GPU loop with 2x360mm radiators, although at the time I ran those tests it might have been 2x240mm, can't remember for sure. EK GPU block, Optimus CPU block. I don't think the GPU temp went over 60C. There seems to be a number of different things that cause these cards to throttle though, not just temp. Possibly throttled based on power? I should try dropping the power target and voltage rather than just putting them all on max. Bit of a juggling act getting everything just right. The GPU memory refuses to go any higher than +1000, so that may have something to do with it.
 
The CPU runs stable under load with 5.3GHz on all cores, so it seems to be a decent chip fortunately. Won't do 5.4 stable though, even on just a couple of cores, which is a bit of a head scratcher.
 
 


  BUILD 1 2   |   MINI-ITX BUILD   |   MODSRIGS $1K WIN   |   HEATWARE 111-0-0   |   ASSOCIATE CODE CSKKXUT5Q9GVAFR

#34
sethleigh
SSC Member
  • Total Posts : 796
  • Reward points : 0
  • Joined: 2015/08/12 11:27:56
  • Status: offline
  • Ribbons : 4
Re: Highest timespy score I can get with 3080 XC3 Ultra 2021/12/29 07:05:18 (permalink)
Nereus
 
Nope.. water cooled; full CPU/GPU loop with 2x360mm radiators, although at the time I ran those tests it might have been 2x240mm, can't remember for sure. EK GPU block, Optimus CPU block. I don't think the GPU temp went over 60C. There seems to be a number of different things that cause these cards to throttle though, not just temp. Possibly throttled based on power? I should try dropping the power target and voltage rather than just putting them all on max. Bit of a juggling act getting everything just right. The GPU memory refuses to go any higher than +1000, so that may have something to do with it.
 
The CPU runs stable under load with 5.3GHz on all cores, so it seems to be a decent chip fortunately. Won't do 5.4 stable though, even on just a couple of cores, which is a bit of a head scratcher.

My guess is going to be thermal throttling of the GPU then. Your CPU cores are hitting higher clockspeeds than mine, but there are 4 fewer of them, so it's no surprise my CPU score is much better. That your graphics score is down ~1k from mine, though, with a 3090 vs 3080ti, is sort of surprising. With your average GPU temp of 51 C I'd bet you are seeing times during the runs where the GPU temp has crossed the threshold for thermal throttling. With my average GPU temp of 36 C during those runs it's insanely unlikely that I ever saw thermal throttling at any point.
 
It's also possible I'm seeing lower water temps due to having more radiator surface area. I've got a single 140mm, a 280x140mm, and a 360x180mm radiator. The Torrent case will support this 180mm wide rad up front, which is actually why I bought the Torrent in the first place (I happened to have a 360x180mm rad). This adds up to around three full 360x120mm rads' worth of surface area.

Happy EVGA customer.  Affiliate Code: 0Y7-1VU-ATW2
 
GigaByte X570 Aorus Master, AMD Ryzen 5900x under Optimus Foundation block, 32gb G.Skill DDR4 @ 3800 MHz 14-14-14-28, EVGA 3080ti FTW3 Ultra under Optimus block, 2TB 980 Pro SSD, EVGA Supernova G6 850W PS, ASUS 34" 3440x1440p 120Hz ultrawide, Lenovo 24" 1080p secondary monitor, Win 10

#35
Nereus
Captain Goodvibes
  • Total Posts : 18916
  • Reward points : 0
  • Joined: 2009/04/09 20:05:53
  • Location: Brooklyn, NYC.
  • Status: offline
  • Ribbons : 58
Re: Highest timespy score I can get with 3080 XC3 Ultra 2021/12/29 07:25:43 (permalink)
sethleigh
Nereus
Nope.. water cooled; full CPU/GPU loop with 2x360mm radiators, although at the time I ran those tests it might have been 2x240mm, can't remember for sure. EK GPU block, Optimus CPU block. I don't think the GPU temp went over 60C. There seems to be a number of different things that cause these cards to throttle though, not just temp. Possibly throttled based on power? I should try dropping the power target and voltage rather than just putting them all on max. Bit of a juggling act getting everything just right. The GPU memory refuses to go any higher than +1000, so that may have something to do with it.
 
The CPU runs stable under load with 5.3GHz on all cores, so it seems to be a decent chip fortunately. Won't do 5.4 stable though, even on just a couple of cores, which is a bit of a head scratcher.

My guess is going to be thermal throttling of the GPU then. Your CPU cores are hitting higher clockspeeds than mine, but there are 4 fewer of them, so it's no surprise my CPU score is much better. That your graphics score is down ~1k from mine, though, with a 3090 vs 3080ti, is sort of surprising. With your average GPU temp of 51 C I'd bet you are seeing times during the runs where the GPU temp has crossed the threshold for thermal throttling. With my average GPU temp of 36 C during those runs it's insanely unlikely that I ever saw thermal throttling at any point.
 
It's also possible I'm seeing lower water temps due to having more radiator surface area. I've got a single 140mm, a 280x140mm, and a 360x180mm radiator. The Torrent case will support this 180mm wide rad up front, which is actually why I bought the Torrent in the first place (I happened to have a 360x180mm rad). This adds up to around three full 360x120mm rads' worth of surface area.

Come to think of it, I may have been running in a smaller case when I did those 3DMark runs too - I'm now in a roomy Corsair 7000X, and with more radiator area. I'm quickly running an AIDA64 GPU stability test as I write this, but under maintained full load for 10 minutes, the GPU has leveled off at 55C, with 57C on the memory... I don't have fans on full though, and ambient room temp is currently 78F. Might give it another 3DMark shot later (and maybe open the window, lol). Not sure why thermal throttling is set to kick in at what I consider a relatively low temp, but I guess it would take a custom bios or firmware to change that. Your average GPU temp at 36C is awesome.
 
Soon as EVGA release the Z690 Dark (and as soon as I can get my mitts on some DDR5 to go with it), my CPU score should jump up a fair bit.
 
post edited by Nereus - 2021/12/29 07:37:35


  BUILD 1 2   |   MINI-ITX BUILD   |   MODSRIGS $1K WIN   |   HEATWARE 111-0-0   |   ASSOCIATE CODE CSKKXUT5Q9GVAFR

#36
sethleigh
SSC Member
  • Total Posts : 796
  • Reward points : 0
  • Joined: 2015/08/12 11:27:56
  • Status: offline
  • Ribbons : 4
Re: Highest timespy score I can get with 3080 XC3 Ultra 2021/12/29 15:35:58 (permalink)
I don't know what the NVidia firmware threshold is for thermal throttling on these newer cards. I believe on the 1080ti the threshold for the first step back in thermal throttling was 55 C.
 
My CPU and GPU temps have always been observed to be some fixed offset from my water temps given the same CPU or GPU load. The amount and quality of the radiators and airflow through them will have a direct bearing on this. One way to reduce the component temps, then, without changing the blocks out for more efficient ones, is to reduce the water temps, and this is most directly done through some combination of more radiator area and/or better/more airflow through the radiators. I like to have all of my radiators set up as intake so that the air going through them is at the ambient room temp and not already pre-heated by other components. With a very open-backed case like the Torrent setting up a positive pressure system like this is awesome, because there's plenty of ways for the air to get out even if every fan in the system is an intake. My singleton 140mm rad fan is an exhaust, though, simply because it's mounted on the rear wall where the rest of the air is all exiting the system.

Happy EVGA customer.  Affiliate Code: 0Y7-1VU-ATW2
 
GigaByte X570 Aorus Master, AMD Ryzen 5900x under Optimus Foundation block, 32gb G.Skill DDR4 @ 3800 MHz 14-14-14-28, EVGA 3080ti FTW3 Ultra under Optimus block, 2TB 980 Pro SSD, EVGA Supernova G6 850W PS, ASUS 34" 3440x1440p 120Hz ultrawide, Lenovo 24" 1080p secondary monitor, Win 10

#37
cotter32
Superclocked Member
  • Total Posts : 231
  • Reward points : 0
  • Joined: 2018/07/19 01:36:14
  • Status: offline
  • Ribbons : 6
Re: Highest timespy score I can get with 3080 XC3 Ultra 2022/01/16 14:39:03 (permalink)
 
I turned some things on, and some off, and got a little better score .
 
Over-all 14284
Graphics 16592
CPU 7989
 
System :
Intel 8086K CPU .
EVGA Z370 Classified K M/B in UFEI mode, above 4G, and resizable bar activated .
G.SKILL TridentZ Series 32GB (4 x 8GB) 288-Pin DDR4 SDRAM DDR4 3200 ( cl 14-14-34 ) .
EVGA GeForce RTX 3080 XC3 ULTRA, 10G-P5-3885KH, ( Video card, in PCIe slot PE2 ) .
GIGABYTE AORUS SSD AIC 1TB PCI-Express 3.0 x4, NVMe 1.3 ( System drive, in PCIe slot PE6, GPT formatted ) .
HyperX Savage 2.5" 240GB SATA III Internal SSD, ( data storage, SATA port 2, GPT formatted ) .
WD Black 1TB Performance Desktop Hard Disk Drive - 7200 RPM, ( data storage, SATA port 3, GPT formatted ) .
LG WH16NS40 Internal SATA 16x Super Multi Blue DVD burner ( in SATA port 1 ) .
Windows 10 home 64 bit operating system, all drives GPT formatted by Windows .
EVGA SuperNOVA P2 Platinum 850W Modular Power Supply ( eco mode off ) .
LG 32” Class QHD Gaming Monitor with G-SYNC 32GK850G-B .
Azio lighted USB keyboard, Logitech USB mouse, Xbox1 USB game controller .
BIOS version 1.18, latest Windows, all drivers up to date, and Microsoft,
or Vendor provided .
 
To the Moon !
#38
Page: < 12 Showing page 2 of 2
Jump to:
  • Back to Mobile