2019 17" HP OMEN 17-cb0065TX (RTX2070) [9th,6C,H] + RTX 3080 @ 32Gbps-TB3 (Razer...
 
Notifications
Clear all

2019 17" HP OMEN 17-cb0065TX (RTX2070) [9th,6C,H] + RTX 3080 @ 32Gbps-TB3 (Razer Core X Chroma) + Win10 1909 // Vive Cosmos Elite  

 of  2
  RSS

Mo oi
(@mo_oi)
Active Member
Joined: 6 months ago
 

I have no room to place large and heavy gaming desktops, so I choose to use notebooks and eGPU.

After my iMac build, Apple announced the end of intel Macs in near future. So, I decided to buy a windows gaming note with the Tb3 port.

While I was playing Fallout4VR, I noticed RTX2060super in eGPU couldn't maintain good performance, so I decided to buy the new GPU.

2 NVIDIA GPUs are so close to conflict each other, I need some tweaks to work them together, especially in VR.

 

[spec]

2019 17" HP OMEN 17-cb0065TX

Intel Core i7-9750H Processor

16GB DDR4

Windows1909

240Hz, full HD internal monitor (in this build, I mainly use this monitor, not the external monitor )

CableMatters' active 1m Tb3 cable

RTX3080 TUF O10G 

[Installation]

DDU the driver and disable auto update. Install driver for RTX 3080 manually, which recognizes the original RTX2070(dGPU). Plug the eGPU and disable RTX2070 and re-install the same driver*1, which recognizes RTX3080.

When 2070 and 3080 are active, playing on screen games assigning high performance to RTX3080 in the graphic settings worked great.

BUT

When playing VR, error 210 occurred which said "composition error". I googled and it said using the different GPUs for the monitor and VR, that error message showed. So I disabled RTX2070 and VR works, but heavy stuttering in many games happened.*2

NEXT

I connect the external monitor and now onscreen games and VR has worked well, which made good 240Hz internal screen useless

FINALLY

I use the headless dummy adaptor to the eGPU HDMI port, which avoided connecting the external monitor. a savior.

[Timespy benchmark]*3

external monitor GPU13544

internal monitor GPU13322

[Comments]

I am very happy with the result because the timespy scores are equal to 2080ti in the desktop setting. Fallout4VR runs well even in the high supersampling setting.

1. I don't know what procedure is correct, but if install the driver when both RTX2070 and 3080 are active, I suppose conflict would happen, so I chose this way. Perhaps at first installing the driver for RTX2070 and after plugging eGPU, installing the driver for RTX3080 may be better?, just overwriting RTX2070 driver?

2. I don't know why. I suppose disabling original dGPU cause whole system instability. I found the same report here

3. No difference between internal and external. This seems odd, but I don't know why. By the way, the same RTX3080 iMac build using external monitor scored GPU14850

 

Please tell me if anyone knows the sophisticated way to work NVIDIA dGPU and NVIDIA eGPU together.

 2020 10 09 23.33.15
 2020 10 09 23.33.30

 

 

This topic was modified 2 weeks ago
2019 17" HP OMEN 17-cb0065TX (RTX2070) [9th,6C,H] + RTX 3080 @ 32Gbps-TB3 (Razer Core X Chroma) + Win10 1909 [build link]  

stonk boi and itsage liked
ReplyQuote
Topic Tags
stonk boi
(@stonk_boi)
Active Member
Joined: 6 days ago
 

@mo_oi, Thanks for sharing your build! I was considering on getting a 3070 as an egpu for my 2020 omen with a 2060. However, when I saw the benchmarks of the 10750H paired with an rtx3090 it somehow performs worse than a built in rtx2060. Could you share your benchmarks with the 3080 VS your built in 2070 as I think it would be interesting if the 3080 could actually perform in a 6 core 9750H.

 

To do: Create my signature with system and expected eGPU configuration information to give context to my posts. I have no builds.

.

ReplyQuote
Lycid
(@lycid)
Active Member
Joined: 1 year ago
 

"I am very happy with the result because the timespy scores are equal to 2080ti in the desktop setting. "

 

Ouch... this is an awful result is it not? Considering a 3080 is supposed to hit 30-40% higher than a 2080ti in performance metrics, this would imply you are getting almost a 30-40% loss in performance compared to desktop. Were you running 1080p? The card isn't really designed for 1080P use, it's designed for 4K use, so people who aren't running at 4K resolution give up most of the performance benefit of the card.

 

The fact that you're also getting same results on internal vs external monitor also seems to imply something is wrong. Hopefully not, as I have a very similar processor (8750H) and was contemplating upgrading from a 2080ti to a 3080. Would be very disappointing to learn if there are little to no real gains for doing so.

This post was modified 2 days ago

Laptop: Razer Blade Advanced 15" w/ GTX 2060, i7 8750H, 16GB RAM
eGPU: Razer Core x Chroma w/ 2080ti


stonk boi liked
ReplyQuote
stonk boi
(@stonk_boi)
Active Member
Joined: 6 days ago
 

@lycid, Do you get any huge improvements from upgrading the 2060 dgpu to a external 2080ti? I'm thinking of getting a 3070 as an egpu paired up with the 10750H but after seeing Jarrod's video I doubt that I could get more performance out of the 3070 than the dedicated 2060.

 

To do: Create my signature with system and expected eGPU configuration information to give context to my posts. I have no builds.

.

ReplyQuote
Mo oi
(@mo_oi)
Active Member
Joined: 6 months ago
 

@lycid, Hi

Now I totally agree with you, I was disappointed with the bad performance in VR games. 

At first I thought 3080 on eGPU came with huge performance boost in VR, but I noticed a slight boost in demanding VR games like Loneecho or Fallout4vr, compared to my previous build RTX2060super eGPU. It couldn't handle the high settings with super sampling 130%, causing many stutters. 

So I did more benchmarks to check the VR performance (my angriness made me pay money 20$ for VR mark)

 

VR mark Cyan room: 10393 (Desktop 2080ti scores around 14000)

https://www.3dmark.com/vrm/51742323

 

VR mark blue room: 4105 (Desktop 2080ti scores around 4500)

https://www.3dmark.com/vrm/51741821

 

I knew the thunderbolt bandwidth and CPU bottlenecking the results, but I wanted more performance boost especially in VR.

 

2019 17" HP OMEN 17-cb0065TX (RTX2070) [9th,6C,H] + RTX 3080 @ 32Gbps-TB3 (Razer Core X Chroma) + Win10 1909 [build link]  

ReplyQuote
Mo oi
(@mo_oi)
Active Member
Joined: 6 months ago
 

@lycid, Hi

Now I totally agree with you, I was disappointed with the bad performance in VR games. 

At first I thought 3080 on eGPU came with huge performance boost in VR, but I noticed a slight boost in demanding VR games like Loneecho or Fallout4vr, compared to my previous build RTX2060super eGPU. It couldn't handle the high settings with super sampling 130%, causing many stutters. 

So I did more benchmarks to check the VR performance (my angriness made me pay money 20$ for VR mark)

VR mark Cyan room: 10393 (Desktop 2080ti scores around 14000)

https://www.3dmark.com/vrm/51742323

VR mark blue room: 4105 (Desktop 2080ti scores around 4500)

https://www.3dmark.com/vrm/51741821

I knew the thunderbolt bandwidth and CPU bottlenecking the results, but I wanted more performance boost especially in VR.

 

@stonk_boi, thank you for your reply

I did more benchmarks using VR mark and I noticed CPU are also bottlenecking results.

Please check my VR mark results above.

I checked my iMac build, using VR mark.

Cyan room scored 11477

https://www.3dmark.com/vrm/51899822?  

Blue room scored 4571 

https://www.3dmark.com/vrm/51899925

 

I notice 10% higher scores than my OMEN 17 build using the same GPU.

The difference may come from the thunderbolt controller chip structure(?), not CPU because OMEN 17's CPU is better.

This post was modified 1 day ago
2019 17" HP OMEN 17-cb0065TX (RTX2070) [9th,6C,H] + RTX 3080 @ 32Gbps-TB3 (Razer Core X Chroma) + Win10 1909 [build link]  

stonk boi liked
ReplyQuote
stonk boi
(@stonk_boi)
Active Member
Joined: 6 days ago
 

@mo_oi, Thanks for your reply!

 

I took a look at your 3080 benchmark in the omen, do you have any benchmarks for your dedicated 2070 though? Also, have you looked at the framerates on games as well?

Maybe fo4 VR with the 2070 dgpu against the 3080 egpu?

 

It would be helpful to find out how much of a difference the dedicated gpu vs the external gpu makes. 

 

 

To do: Create my signature with system and expected eGPU configuration information to give context to my posts. I have no builds.

.

ReplyQuote
Mo oi
(@mo_oi)
Active Member
Joined: 6 months ago
 

@stonk_boi

 

Here are dGPU 2070 benchmark

 

Timespy 7529

https://www.3dmark.com/3dm/51907463

 

VRmark cyan 7207

https://www.3dmark.com/vrm/51907568?

 

VRmark blue 2358

https://www.3dmark.com/vrm/51907890?

 

Fallout4VR

dGPU RTX2070

in the complex town like Diamond city, the GPU frame time shows 20-25ms in front of the noodle shop(My monitoring place), super sampling 130%, using FPSVR. I installed some mods optimizing FPS. A lot of stuttering and reprojection. Its unplayable. 

 

eGPU 3080

in the same place, the GPU frame time shows 15-20ms, cannot reach to 90Hz(50-70Hz? <11Hz=90Hz) but I feel a little stuttering and its playable for me.

 

This post was modified 1 day ago
2019 17" HP OMEN 17-cb0065TX (RTX2070) [9th,6C,H] + RTX 3080 @ 32Gbps-TB3 (Razer Core X Chroma) + Win10 1909 [build link]  

stonk boi liked
ReplyQuote
Lycid
(@lycid)
Active Member
Joined: 1 year ago
 

@stonk_boi

 

Yes, I do - but don't expect a miracle. My timespy scores went from high 6-7000's to over 10K IIRC, and VR is much smoother. I run at 1440p on an external. The 2060 actually performs for some games on the same max settings pretty close in FPS at 1080p on the internal monitor to the 2080ti at 1440p on the external monitor - like Borderlands 3 or Apex. Other games I clearly gain bigger performance boost on the 2080ti at 1440p compared to the 1080p internal monitor results from my 2060.

 

The biggest advantage though is temp controls. Laptop run ice cool with the egpu which helps with performance stability.

 

Laptop: Razer Blade Advanced 15" w/ GTX 2060, i7 8750H, 16GB RAM
eGPU: Razer Core x Chroma w/ 2080ti


stonk boi liked
ReplyQuote
Lycid
(@lycid)
Active Member
Joined: 1 year ago
 

@mo_oi

 

Yes, this is likely because the mac has the thunderbolt controller integrated into the CPU. Check the icelake thread for laptops that do the same thing on PC - they get as much as 40% gains over non-icelake CPU's that don't integrate the controller on something like the 3090 which really pushed the TB3 port.

 

Could you do any 4K comparisons or benchmarking? AFAIK thunderbolt & 3080 performance is more optimized for 4K resolution.

 

This post was modified 23 hours ago

Laptop: Razer Blade Advanced 15" w/ GTX 2060, i7 8750H, 16GB RAM
eGPU: Razer Core x Chroma w/ 2080ti


ReplyQuote
 of  2