Need advice on Intel Arc video cards.

[wade7575]

I'm looking at getting an Intel Arc A310 or A380 video card to use with Handbrake because I use Handbrake a lot and I seen a few video's on youtube by guys that seemed credible that using an Intel card will boost your speeds when using Handbrake and other video processing software which is what I want,I won't be using the card for gaming so I'm not concerned about that just getting more processing power.

Have you guys had much to do with the Intel cards and if so how did it work running Handbrake,these are the 2 cards I'm looking at.

https://www.amazon.ca/ASRock-A310-P...id=1716283452&sprefix=intel+arc,aps,89&sr=8-2

https://www.amazon.ca/Asrock-Challe...id=1716283452&sprefix=intel+arc,aps,89&sr=8-1

Another question I have is that I use a GTX960 and GTX570 card in my 2 rig's because I don't really game and I needed a card because my one cpu didn't have built in graphic's.

What the problem I run into is that when I try and use the HDMI mini port no matter what setting I select in Windows 10 or 11 will fit my monitor correctly unless I use the DVI port.

The monitor that I have is a Viewsonic VX2439VM and it's an oldie but a goodie and has always worked great,will I run into the same problem with an Intel Arc card because they have HDMI port's only.

https://www.viewsonic.com/uk/products/sheet/VX2439wm

Post magically merged: May 21, 2024

This may sound really stupid but I was just looking at my monitor and I think I need to set the video input to HDMI,I will have to give that a try.

I have always ran DVI and never thought of changing that setting.

 

[EarthDog]

Not sure I can help on the Arc Handbrake performance... maybe look up some reviews and see. An initial google on my end brought up some AV1 issues (slow), but not sure if that has been changed or not. Hopefully someone can swing by and help out.

 

[JLK03F150]

I know the Intel graphic cards want to have modern motherboard features like gen 4 PCIe and Resizable BAR. If the X99-A doesn't support ReBar it will not perform at its best potential for you.

 

[mackerel]

I'm looking at getting an Intel Arc A310 or A380 video card to use with Handbrake because I use Handbrake a lot and I seen a few video's on youtube by guys that seemed credible that using an Intel card will boost your speeds when using Handbrake and other video processing software which is what I want,I won't be using the card for gaming so I'm not concerned about that just getting more processing power.

Have you guys had much to do with the Intel cards and if so how did it work running Handbrake,these are the 2 cards I'm looking at.

I do have an a380 but not tested it for video encode. I could run a bench on it vs nvidia if you have particular settings you'd like tested. Or quite likely someone else would have done so already if you look for it.

I know the Intel graphic cards want to have modern motherboard features like gen 4 PCIe and Resizable BAR. If the X99-A doesn't support ReBar it will not perform at its best potential for you.

That's for gaming/3D performance. Don't think it matters if only for video encoding. There is a project out there that aims to add ReBAR support to systems that don't already have it. It involves injecting a module into UEFI. I wasn't motivated enough to go through it.

 

[Flamethrower1972]

I don't want to be a Debbie downer, but I did some reading as well as benchmarks tests and the cards do not perform all that well. Did you try to find a Video or something on a goggle search to find out more info to do tests using Handbrake?

 

[wade7575]

I don't want to be a Debbie downer, but I did some reading as well as benchmarks tests and the cards do not perform all that well. Did you try to find a Video or something on a goggle search to find out more info to do tests using Handbrake?

From what I read to they don't but that has something to with realtime crap or something like that in Topaz software,a few guys said with Handbrake and other encoding software it performs 18 to 25% faster going by their test.

Post magically merged: May 21, 2024

I do have an a380 but not tested it for video encode. I could run a bench on it vs nvidia if you have particular settings you'd like tested. Or quite likely someone else would have done so already if you look for it.


That's for gaming/3D performance. Don't think it matters if only for video encoding. There is a project out there that aims to add ReBAR support to systems that don't already have it. It involves injecting a module into UEFI. I wasn't motivated enough to go through it.

If you don't mind try encoding something that is MP4 or MKV at 720x404 settings and see how fast it goes in Handbrake,I do a lot in that size to keep the file size small.

 

[mackerel]

I got some results but I'm not sure how useful it will be as I'm unsure about settings in Handbrake.

Test file is 14:20 duration 1280x720 30fps game content.

I initially picked Fast 480p preset, ticked the constant framerate box (default is peak?) and adjusted encoder.

System 1: 7800X3D + 4070
H264 NVEnc 30s
AV1 NVEnc 30s

System 2: 12100F + A380
H264 IntelQSV 62s
AV1 IntelQSV 60s
CPU usage ~100%

I thought something might be happening here, so I looked at CPU usage. The 2nd system was maxed out so this was a CPU test, not an encoder test.

I tried adjusting the settings to see if I can lower CPU usage. In the end I chose Fast 720p preset (so no resizing needed), manually disabled all filters, removed audio track, and again ticked the constant framerate box.

System 1: 7800X3D + 4070
H264 NVEnc 24s
AV1 NVEnc 25s
CPU usage ~20%

System 2: 12100F + A380
H264 IntelQSV 39s
AV1 IntelQSV 46s
CPU usage ~60%

I can't say I totally removed the CPU element but it does look like the A380 is still slower than the 4070.

Maybe if I rendered to a higher resolution output it will be more encoder limiting than CPU limiting?

Edit: Tom's Hardware testing suggests the results are closer between A380 and NV GPUs: https://www.tomshardware.com/reviews/intel-arc-a380-review/5

 

[EarthDog]

The reality is, almost any GPU he buys that's more modern will be A LOT better than the ancient things he's running on. So while they may be 'slow', that's very myopic advice as it's going to be a lot better than what he has. Exactly how much... no idea, but generations newer should bring a lot better performance.

 

[wade7575]

@mackerel Thanks for the help those result's helped me a bit.

With Handbrake it pretty much runs your cpu at 100% but it not like when you cinebench that truly pounds the crap out of your hardware and push's it to the max truly.

What I was wondering is if you take a look and see what fps Handbrake is processing the video at,with my GTX 960 I'm getting around 440 to 500 fps as it goes up and down and holds more around the 440 fps mark or so.

I'd love to get to 1K fps processing speed but even with the best arc card I better that's a pipe dream but even to get to 600 fps or more would still help shave time off for me.

 

[mackerel]

I gave the source video length and frame rate, so we know how many frames were present: 25800. Might be a tiny error as it would be rounded to nearest second. I only noted the encoding time as that was more easily seen than the fps, which as you noted fluctuated somewhat. I think the main take away here is that for low resolutions, the encoder doesn't seem to be the limit but more the CPU. A faster CPU might help more than a faster GPU, but it will depend on the exact settings.

Fast 480p preset, ticked the constant framerate box and adjusted encoder.

System 1: 7800X3D + 4070
H264 NVEnc 860 fps
AV1 NVEnc 860 fps

System 2: 12100F + A380
H264 IntelQSV 430 fps
AV1 IntelQSV 416 fps


Fast 720p preset (so no resizing needed), manually disabled all filters, removed audio track, and again ticked the constant framerate box.

System 1: 7800X3D + 4070
H264 NVEnc 1075 fps
AV1 NVEnc 1032 fps

System 2: 12100F + A380
H264 IntelQSV 662 fps
AV1 IntelQSV 561 fps

 

[Flamethrower1972]

I gave the source video length and frame rate, so we know how many frames were present: 25800. Might be a tiny error as it would be rounded to nearest second. I only noted the encoding time as that was more easily seen than the fps, which as you noted fluctuated somewhat. I think the main take away here is that for low resolutions, the encoder doesn't seem to be the limit but more the CPU. A faster CPU might help more than a faster GPU, but it will depend on the exact settings.

Fast 480p preset, ticked the constant framerate box and adjusted encoder.

System 1: 7800X3D + 4070
H264 NVEnc 860 fps
AV1 NVEnc 860 fps

System 2: 12100F + A380
H264 IntelQSV 430 fps
AV1 IntelQSV 416 fps


Fast 720p preset (so no resizing needed), manually disabled all filters, removed audio track, and again ticked the constant framerate box.

System 1: 7800X3D + 4070
H264 NVEnc 1075 fps
AV1 NVEnc 1032 fps

System 2: 12100F + A380
H264 IntelQSV 662 fps
AV1 IntelQSV 561 fps

It looks like the 4070 just blows the A380 out of the water IMO.

 

[mackerel]

It looks like the 4070 just blows the A380 out of the water IMO.

The CPU is not the same. I saw significant CPU load. That's why I'm suggesting it could be a CPU limit. I do not have the time or motivation to swap GPUs so the same CPU is used to check this. Note in my 2nd set of testing where I tried to reduce CPU load, the gap narrowed. The other testing I linked done by Tom's Hardware, using higher solutions, showed the gap to be narrower still. I think a high resolution would be a bigger load and shift more onto the GPU than CPU than in my testing.