Shadow of the Tomb Raider benchmark

[phanbuey]

Welp it's no 5800x but not bad for a $390 budget option For whatever reason im 43% GPU bound while @Felix123BU is 0% .

 

[Felix123BU]

View attachment 184754


Welp it's no 5800x but not bad for a $390 budget option For whatever reason im 43% GPU bound while @Felix123BU is 0% .

I am confused by what the benchmark in this game is reporting Sometimes im 0% GPU bound, other times 27%
The Cpu Render section always shows better on Intel, the CPU Game section is better on mine but the CPU Render is quite a lot lower.
I have no clue how the average is calculated, its not the average displayed in the sub section, nor a mathematical division of any of the values...

And btw, the 10850K is a very good processor, I would have considered it have I not have already been on AM4 with a good motherboard. I bet if you overclock the crap out of it it will match my 5800X rather easy, though the 3080 is a smidge weaker at 1080

 

[RandallFlagg]

Kinda surprised I have not seen a 10600K in this thread.

 

[mouacyk]

Welp it's no 5800x but not bad for a $390 budget option For whatever reason im 43% GPU bound while @Felix123BU is 0% .

3080 is dozing off at 1080p, no fault of CPU.

 

[nguyen]

some 3090 loving

 

[mrthanhnguyen]

some 3090 loving

View attachment 184920

U run stock? Coz my system has like 70-80 more fps.

 

[nguyen]

U run stock? Coz my system has like 70-80 more fps.

I'm pretty sure not at 4K

 

[mrthanhnguyen]

I'm pretty sure not at 4K

Oh sorry thought it 1440p.

 

[mouacyk]

4k, dsr on 2560x1440 monitor

 

[Felix123BU]

4k, dsr on 2560x1440 monitor
View attachment 185023

what's the difference in quality vs native 2k?

 

[mouacyk]

what's the difference in quality vs native 2k?

Much sharper image due to the downsampling, but obviously not as sharp as native 4K monitor.

 

[Lew Zealand]

Intel NUC8i5 4c8t @3.6GHz ACT, 16GB 2400 MHz, GTX 1060-6GB in Akitio Node eGPU.

FPS seems OK compared to other 1060s in 4c8t machines though the min and 95% are a bit low as expected from TB3 latency.

Core i5-8400, B360, 16GB 2666MHz CL13-13-13-35, PNY GTX 1080 w/crap cooler, so UV to .95 and OC to roughly 2012-2025MHz, temp dependent.

If I get real frisky, I'll swap the GPUs and test vice versa to see how the eGPU affects them differently. Yeah. We'll see if I do that.

 

[Lew Zealand]

Got around to swapping the GPUs. Only the NUC w/1080 for now:

Intel NUC8i5 4c8t @3.6GHz ACT, 16GB 2400 MHz, PNY GTX 1080 w/crap cooler (UV to .95 and OC to roughly 2012-2025MHz, temp dependent) in Akitio Node eGPU.

Clear CPU limitation. Ran it at 1440p for fun and got 62fps, 92% GPU-bound, with the CPU bound spots being NPC-heavy. The 1080 is a better 1440p card than the 1060 is a 1080p card, not surprising with twice the CUDA cores.

1060 in 8400 to come later.

 

[mrthanhnguyen]

1 more mora so 1 more update
1440p

4k

 

[Lew Zealand]

Finally, the 1060 in the i5-8400 box:

Core i5-8400, B360, 16GB 2666MHz CL13-13-13-35, MSI Gaming X GTX 1060-6GB

15-16% hit in overall FPS at 1080p when swapping the 1060 from a gaming desktop to an eGPU. Not too bad. But a 20% hit to 1% lows as you might expect with the reduced throughput and added latency of PCIe-TB3-PCIe translation.

Compare to 23-24% hit in overall FPS at 1080p when swapping the 1080 from a gaming desktop to an eGPU. However the difference here is that the 4c8t CPU in the NUC now becomes the limitation in NPC-heavy areas with a 1080, which leads to about 37% reduction in 1% lows. If I had the 6c12t NUC10i7, that would give a more apples-to-apples comparison but yeah, that one's pretty expensive...

OK and I've played a few games on the 1080 in the NUC+eGPU compared to the gaming PC i5-8400 and here's what I see:

Ark:SE - model pop-in is bad in this game in general and is noticeably worse in the eGPU with obvious frame drops. However most of the time, it plays pretty similarly to the gaming desktop with only a slightly lower FPS at 1440p with my medium-high custom settings.

Control - Notably worse FPS but no pop-in problems. This game is pretty GPU-heavy yet easy on the CPU, but they must "talk" a lot as the average FPS is notably lower at 1440p Medium, like ~53fps in the gaming PC to ~35 fps in the NUC+eGPU. I did not expect this.

DiRT Rally - FPS a little lower at 1440p Ultra yet still high enough not to be noticeable, feels the same. No pop-in problems though the FPS in the first 10 seconds of a race are notably lower - texture load-ins? but no dropped frames, just noticeably lower overall FPS at that time which is then alleviated in about 10 sec.

Of course I haven't actually *played* SotTR yet as I'm still working on RotTR but I haven't played that on the NUC+eGPU yet. Maybe tonight.

 

[Keullo-e]

Highest but RT shadows on @ medium

 

[Fizban]

11% GPU bound, nice cpu benchmark, lul

 

[harm9963]

1080Ti still has life
My 5950X elevated my 1080Ti for sure !

 

[Keullo-e]

1080Ti still has life

Exactly.

 

[Fizban]

1080 TI's a solid gpu still, but the cpu is doing a lot of work there, the game is very cpu-bound usually.

 

[Keullo-e]

1080 TI's a solid gpu still, but the cpu is doing a lot of work there, the game is very cpu-bound usually.

I feel pretty balanced with an AMD 3600

 

[Fizban]

I feel pretty balanced with an AMD 3600

My i7 holds back my laptops GPU hard. I'd have to compare at 1440P or 4K to beat his 1080 TI as a result.

 

[Keullo-e]

My i7 holds back my laptops GPU hard. I'd have to compare at 1440P or 4K to beat his 1080 TI as a result.

Anyway, best purchase ever as I paid 300EUR from this last week

 

[thesmokingman]

Just ran this on a G9, stock with maxed power.

1080

5120

 

[QuietBob]

Haven't seen a 3300X here, so here ya go. DX 12 @ 1080p, highest preset:



Yes, I'm aware of the GPU bottleneck This is a 9 year old GPU. I'd still say the game is playable at around 30 fps.
But check out that little beast of a CPU with 4c/8t. Even an RTX2080 wouldn't be bottlenecked at these settings.

EDIT: Looking at the scores above, it could also easily drive an RTX3090 at 4K