[Clukos]

Those are some great scores with 370W!

I think I'm getting away with it due to the very good memory chips I got on this 3080

The clock range is nothing special but not bad either, it usually hovers around 2.1GHz in benchmarks.

I wish there was a way to unlock the power without shunt mod, I sold my 1080 Ti and if I manage to kill the 3080 it'll be a long time before I can find another one at MSRP :/

 

[BluePaint]

I wish there was a way to unlock the power without shunt mod

Indeed! I skipped last generation. Was there such a thing for Volta?

 

[Imprezzion]

Yup, 2080 Ti had several unlocked BIOS. I ran the Kingpin XOC 2000w one on a reference PCB A chip. I saw 524w in Superposition 4K Optimized and game power draw was about 440-470w @ 2160Mhz 1.125v.

 

[BluePaint]

@Clukos
What did modify on your FE?
Managed to squeeze out 12635 points in PR from unmodified FE
100% fans + Window Mod = 40C @ + 1300 VRAM @ + 250 core = 2160max, 2110 avg

 

[edhutner]

What should be expected max stable core oc for siprim x 3080 on water? And what is best method for stability testing.

Currently in MSI AB I have voltage+100, power 116%, core+110mhz. It is stable in benches and gaming. Additionally I have run couple of hours MSI kombustor with artifact scanner.

Assetto Corsa Competizione is the game I play mostly and there I get about 2055-2070mhz and voltage limited at max 1.087v.

I have tried AB oc scanner. It made curve average +70 or +90mhz (dont remember). This average result is not very encouraging. And that makes me think that my manual +110mhz may surprise me with crash in some moment.

I want to go higher, but I want to be sure that it will be 100% stable. I dont want to crash during online race in ACC. This is why I nees kind of more emperical method to validate stability.

 

[BluePaint]

If stability is of most importance, just be conservative and back off your OC by 40 or 50Mhz and be safe. Don't think you will notice 1-2% less performance. Trying to squeeze out the last 1 or 2 % and going for absolute stability at the same time doesn't go together.

I would also cap framerate which in addition to taxing the GPU less on average gives usually smoother gameplay if choosen correct (somewhat lower than peak framerate for most of the time).

 

[Imprezzion]

Probably around 2130 ish. It's a lottery. Only way to find out is to test it lol. I fixed my crashes in Control somehow. It's related to temperatures appearantly. On like, 65% fan speed with the core around 62c, hotspot 74c junction 86c it crashes with +105 but when I crank the fans to 100% with temps of core 55c hotspot 63c junction 74c it runs fine at +105 for hours. So, these cards do seem to be just as temperature sensitive as RTX2xxx was in some scenarios.

Still, even at max power limit 370w and with no fans or RGB power from the card it still limits hard in Control with no DLSS, 1080p, all max quality, 4x MSAA, RT maxed as well. At +105 it barely holds it above 1900Mhz with effective clocks sometimes even going under it to like ~1890Mhz and voltage sits around 0.943-0.962v most of the time. Performance is outstanding on the other hand, it never drops below 80 FPS and averages like 95 FPS in that game and it's unbelievably smooth with Ultra Low Latency mode enabled.

If I run a lighter game like BF4 @ 288 FPS cap (2x144) or world of tanks with 144Hz G-Sync enabled it runs around 2070-2085Mhz all the time at 1.087-1.100v and never hits power limits either. Effective clocks 2060-2080Mhz all the time.

 

[BluePaint]

Using last cold days with 0C air:
2nd Place FS-Ultra-GPU Leaderboard 12730
2nd Place FS-Extreme-GPU Leaderboard 25029

 

[Clukos]

@Clukos
What did modify on your FE?
Managed to squeeze out 12635 points in PR from unmodified FE
100% fans + Window Mod = 40C @ + 1300 VRAM @ + 250 core = 2160max, 2110 avg

I have the EK waterblock on it with liquid metal, which allowed me to run a bit higher sustained clocks, I also got +1900 on the mem (that's the point where it stops giving me extra performance). Otherwise, I think the main limiting factor is the power limit atm.

 

[arrow0309]

Hiya, just bought a Bykski block for my 3080 Trio from AliExpress, this one exactly (without the bp anyway):

Bykski Full Coverage GPU Water Block and Backplate for MSI RTX 3080/3090 GAMING X TRIO (N-MS3090TRIO-X)

The Bykski Full Coverage GPU Water Block for MSI RTX 3080/3090 GAMING X TRIO is a work of art! This full coverage block directly cools the video cards core (GPU), graphics memory (RAM), and voltage regulator module (VRM). By directly cooling these components you can achieve great temperatures...

www.bykski.us

And I was wondering if someone can post / link a manual, I just want to see the thermal pads like dimensions, especially those 2 slim stripes on the mosfets.
Or if anyone used it, I'd like to know if they were 0.5mm or 1mm thickness.
So I may order better ones.
On the memory ic's and / or some other coils / poscaps zones they should probably be like 1mm, I can use those bundled or find something inside my house.
But I'd like to provide better ones for the mosfet, I don't know what are they bundling with this cheap block (paid £85, taxes included).
Thanks

 

[jura11]

@arrow0309

Bykski using 1.2-1.25mm thermal pads, their w/mk rating I'm not sure but looks like they're 4w/mk

Running Bykski waterblocks on my two RTX 3090 GamingPro and no issues, temperatures are great in 36-38°C in gaming and in rendering temperatures are in 30's (32-35°C), running on both RTX 3090's KPE XOC BIOS with power limit set to 65% and VRAM temperatures won't break 60°C on both GPUs with 1495MHz OC on VRAM and on other 1200MHz OC on VRAM

I have built loop for friend with 10900k and RTX 3090 and we used 360mm radiator on top and 240mm radiator on bottom and in his loop we have seen 42-45°C as highest temperature on RTX 3090 with KFA2 390W BIOS

If you will gain something running better thermal pads like Thermalright Odyssey, I have no idea because I didn't tried yet better pads on my GPUs

Hope this helps

Thanks, Jura

 

[arrow0309]

@arrow0309

Bykski using 1.2-1.25mm thermal pads, their w/mk rating I'm not sure but looks like they're 4w/mk

Running Bykski waterblocks on my two RTX 3090 GamingPro and no issues, temperatures are great in 36-38°C in gaming and in rendering temperatures are in 30's (32-35°C), running on both RTX 3090's KPE XOC BIOS with power limit set to 65% and VRAM temperatures won't break 60°C on both GPUs with 1495MHz OC on VRAM and on other 1200MHz OC on VRAM

I have built loop for friend with 10900k and RTX 3090 and we used 360mm radiator on top and 240mm radiator on bottom and in his loop we have seen 42-45°C as highest temperature on RTX 3090 with KFA2 390W BIOS

If you will gain something running better thermal pads like Thermalright Odyssey, I have no idea because I didn't tried yet better pads on my GPUs

Hope this helps

Thanks, Jura

Hello mate, thank you for the comprehensive reply.
It's OK then, I'll stick to those bundled in that case, no worries.
One last question, have you applied them only to the VRAM and to the 2 main mosfet stripes (like they're indicating) or did you put 2 more on the chokes and the poscaps (left side only)?

 

[mouacyk]

@jura11 and @arrow0309 Doesn't the pad thickness depend on who made the card? Palit may not have the same height offset as MSI when it comes to VRAM and VRMs. For example, my card is Gigabyte Eagle OC and all pads are 1mm, the cheapo 2-3 W/mk type.

 

[jura11]

Hello mate, thank you for the comprehensive reply.
It's OK then, I'll stick to those bundled in that case, no worries.
One last question, have you applied them only to the VRAM and to the 2 main mosfet stripes (like they're indicating) or did you put 2 more on the chokes and the poscaps (left side only)?

Hi there

No worries there, yes I have applied thermal pads only on indicated places(2 main mosfet stripes and VRAM and I placed them too on POSCAP), no I didn't put them on the chokes and not sure if that would make any difference in temperatures

In something like mining VRAM temperatures would be off course with 1200MHz OC VRAM higher, in my case I have seen something around 72°C as max on top GPU and on bottom at 82°C, that's VRAM temperatures and core temperatures been in 32-36°C with 105MHz

Hope this helps

Thanks, Jura

 

[arrow0309]

@jura11 and @arrow0309 Doesn't the pad thickness depend on who made the card? Palit may not have the same height offset as MSI when it comes to VRAM and VRMs. For example, my card is Gigabyte Eagle OC and all pads are 1mm, the cheapo 2-3 W/mk type.

Not exactly since the waterblock is also custom product and can be made with identical or different offsets.

Hi there

No worries there, yes I have applied thermal pads only on indicated places(2 main mosfet stripes and VRAM and I placed them too on POSCAP), no I didn't put them on the chokes and not sure if that would make any difference in temperatures

In something like mining VRAM temperatures would be off course with 1200MHz OC VRAM higher, in my case I have seen something around 72°C as max on top GPU and on bottom at 82°C, that's VRAM temperatures and core temperatures been in 32-36°C with 105MHz

Hope this helps

Thanks, Jura

Jeez, kinda high temps (for a water cooled hw) but also your vrams are rocking pretty high.
Mine are like 10k daily and a +105-120 on the gpu, also being a 3080 with less memory ic's should ensure lower temps all around.
So yeah, maybe I'll stick with their bundled pads as well.
Cheers mate

 

[jura11]

Not exactly since the waterblock is also custom product and can be made with identical or different offsets.


Jeez, kinda high temps (for a water cooled hw) but also your vrams are rocking pretty high.
Mine are like 10k daily and a +105-120 on the gpu, also being a 3080 with less memory ic's should ensure lower temps all around.
So yeah, maybe I'll stick with their bundled pads as well.
Cheers mate

Hi there

I normal gaming I see 36-38°C and in rendering I see 32-33°C as max and VRAM temperatures in gaming won't break 60°C and in rendering they're in 60's all the time

I'm running 1495MHz on VRAM on first one and on second one I use 1200MHz like for rendering or benchmarks

Only in mining you see higher VRAM temperatures, normally hotspot is in low 40's and core temperatures are in 30's with KPE XOC BIOS capped at 65%(with that BIOS capped at 65% GPU pulls like 446W each)

Hope this helps

Thanks, Jura

 

[Hiikeri]

I get my MSI Suprim X 3080 yesterday. With stock air cooler + stock bios.

-3dMark Time Spy graphics score > 20.2k
-3dMark Port Royale > 13k

I scored 18 042 in Time Spy

Intel Core i9-9900K Processor, NVIDIA GeForce RTX 3080 x 1, 32768 MB, 64-bit Windows 10}

www.3dmark.com

I scored 13 033 in Port Royal

Intel Core i9-9900K Processor, NVIDIA GeForce RTX 3080 x 1, 32768 MB, 64-bit Windows 10}

www.3dmark.com

 

[BluePaint]

@Hiikeri
Looks like good chip!

Average clock speed in benchmarks and the temps look very similar to my Trio with Strixx BIOS and air cooler
Your [email protected] Ghz and > 4000Mhz RAM is worth a lot of points

 

[edhutner]

Wow great scores. I can not get these with my 3080 suprimx and it is watercooled. My last results are timespy graphics score - 19355, port royal - 12544.

Also I note that my gpu temperature is kind of wrong. Especially when idling it is sometimes shown a 1-2 degrees lower than my coolant temperature, and that is impossible.

 

[Hiikeri]

We are now investigating here in FInland at strange bios issue what i found about 2h ago.

Those my original SuprimX results, on my TimeSpy average Mhz was 2110Mhz

And my PortRoyal bench my average Mhz was 2123Mhz.

But whats a strange and issue now, i flashed my MSI card to Asus StrixOC 450W bios...

And i can increase my TimeSpy and PortRoyal average Mhz, AVG!!! MHZ about +50Mhz, but my 3dMark points are lower!!!

My PortRoyal avg. Mhz on now StrixOC 450W bios are even 2170Mhz, but my PR scores are still lower than my SuprimX bios @ avg. 2123Mhz

My VRAM speed was throttling on my SuprimX scores, but not on my ASUS bios results (Max.mem.clock +1375, avg. also +1375).

I think, and we are investigating this issue here on Finland, i get +50Mhz more GPU but my scores are still lower than my original MSI SuprimX (430W) bios...

At 1st sight, many of ours 3080 owners think that its maybe somekind ASUS cheat-Mhz issue, higher clocks without more power!!!

That could be explanation why most of youtube reviewers can drive Strix cards so high clocks. Usually SuprimX, FTW3, GB xTreme or Galax & Top-Tier cards can get only 2075-2100Mhz, and always StrixOC cards can get at same situation ~2135-2150Mhz! At least they reviewers run on theirs videos TombRaider... you know... On Asus cards theirs clocks are about +50Mhz higher than another brands 3080.

I scored 13 033 in Port Royal < avg.2123Mhz SuprimX 430W bios, 13033 scores (Vram +1366Mhz, slight throttling)
I scored 13 024 in Port Royal < avg.2170Mhz ASUS 450W bios, 13024 scores (Vram +1375Mhz, NO throttling)