5950x and Handbrake

dpriest

Limp Gawd
Joined
Nov 18, 2014
Messages
131
It seems the 5950x is a gamer's dream and most people grabbing this processor are building gaming rigs. All of the benchmarks I have seen regarding video transcoding have been compared to other Ryzen CPUs and Intel chips but I have not seen one benchmark comparing the Handbrake performance of the 5950x to Threadripper processors. I find that strange that maybe AMD specifically doesn't want the 5950x compared to Threadripper performance maybe because it will sway people away from Threadripper because of the high cost.

Has anyone experienced themselves or seen Handbrake performance comparisons between the 5950x and Threadripper?
 
It seems the 5950x is a gamer's dream and most people grabbing this processor are building gaming rigs. All of the benchmarks I have seen regarding video transcoding have been compared to other Ryzen CPUs and Intel chips but I have not seen one benchmark comparing the Handbrake performance of the 5950x to Threadripper processors. I find that strange that maybe AMD specifically doesn't want the 5950x compared to Threadripper performance maybe because it will sway people away from Threadripper because of the high cost.

Has anyone experienced themselves or seen Handbrake performance comparisons between the 5950x and Threadripper?

I'd have done that very comparison if AMD would have sampled either Threadripper or a 5950X to us for testing. The latter we'll be acquiring eventually, but the Threadripper at present isn't something I'll likely test.
 
I'd have done that very comparison if AMD would have sampled either Threadripper or a 5950X to us for testing. The latter we'll be acquiring eventually, but the Threadripper at present isn't something I'll likely test.
If you want, I could feed it through my Threadripper system - 3960X, just to give you example numbers. Just tell me what presets you want.
 
If you want, I could feed it through my Threadripper system - 3960X, just to give you example numbers. Just tell me what presets you want.
That's cool of you to offer. I don't have the 5950x. Maybe I started a thread where we can do our own benchmarks. I built a system with the Threadripper 1950x and am thinking about upgrading if there is a sizeable improvement in performance. I can provide a 20 minute 1080P video clip and share it with others. I use very high quality encoding settings in Handbrake. If we all set Handbrake with the same settings and process the same video we have apples to apples comparison of FPS and time taken to encode. If anyone has the 5950x that would be great also.

Let me know who is game.

Thanks!
 
That's cool of you to offer. I don't have the 5950x. Maybe I started a thread where we can do our own benchmarks. I built a system with the Threadripper 1950x and am thinking about upgrading if there is a sizeable improvement in performance. I can provide a 20 minute 1080P video clip and share it with others. I use very high quality encoding settings in Handbrake. If we all set Handbrake with the same settings and process the same video we have apples to apples comparison of FPS and time taken to encode. If anyone has the 5950x that would be great also.

Let me know who is game.

Thanks!

There would be a massive difference between a Threadripper 1950X and a 3960X. Zen 2 was considerably faster anyway, but you also have better clock speeds and a much higher CPU core / thread count. The platform has improved, it can clock memory higher, etc.
 
There would be a massive difference between a Threadripper 1950X and a 3960X. Zen 2 was considerably faster anyway, but you also have better clock speeds and a much higher CPU core / thread count. The platform has improved, it can clock memory higher, etc.
Several months back someone on this forum processed a video I provided through a 3950x system even though the 3950x was faster the performance increase did not justify the expenditure on a new MOBO, processor and cooler.
 
Several months back someone on this forum processed a video I provided through a 3950x system even though the 3950x was faster the performance increase did not justify the expenditure on a new MOBO, processor and cooler.

A 3950X isn't a 3960X. As for what's worth the expenditure, that's all relative. For gaming, I'd upgrade this 3950X to a 5950X without a moments hesitation. There are a lot of people who wouldn't do that.
 
A 3950X isn't a 3960X. As for what's worth the expenditure, that's all relative. For gaming, I'd upgrade this 3950X to a 5950X without a moments hesitation. There are a lot of people who wouldn't do that.
I'm planning on jumping from the 3960X to a 5960X when it arrives too. Also frees up the 3960X to replace the 1950X I have in the HTPC - just need to find another TRX40 board. The 1950 is twitchy.
 
Has anyone experienced themselves or seen Handbrake performance comparisons between the 5950x and Threadripper?

In multi threaded environment cores over clocks still reigns supreme. You can see this in the cinebench runs. While the 5950x is vastly improved it still is far away from a 24-32 core threadripper 3. For ex in R20 5950x is around 11K in multi where the 3960x is at 14K and 3970x at 17K. And to put it into some clock frequency perspective, a 5950x at 6.3ghz on LN2 scores gets 15k.

That said the 5950x is very compelling for a small but dense render station but it is no replacement for the sheer core count of threadripper.
 
A 3950X isn't a 3960X. As for what's worth the expenditure, that's all relative. For gaming, I'd upgrade this 3950X to a 5950X without a moments hesitation. There are a lot of people who wouldn't do that.
I'm specifically interested in seeing what the 5950x has to offer in Handbrake compared to the TR 1950X. The TR 3970X is SLIGHTLY faster than the 3950X in Handbrake. I have eliminated the 3000 series Threadrippers and 3950X because again specifically with Handbrake the performance increase does not justify the expenditure for me.
 
I'm specifically interested in seeing what the 5950x has to offer in Handbrake compared to the TR 1950X. The TR 3970X is SLIGHTLY faster than the 3950X in Handbrake. I have eliminated the 3000 series Threadrippers and 3950X because again specifically with Handbrake the performance increase does not justify the expenditure for me.

Handbrake isn't great for cpu. You should look at using staxrip which can take advantage of massive core counts.
 
In multi threaded environment cores over clocks still reigns supreme. You can see this in the cinebench runs. While the 5950x is vastly improved it still is far away from a 24-32 core threadripper 3. For ex in R20 5950x is around 11K in multi where the 3960x is at 14K and 3970x at 17K. And to put it into some clock frequency perspective, a 5950x at 6.3ghz on LN2 scores gets 15k.

That said the 5950x is very compelling for a small but dense render station but it is no replacement for the sheer core count of threadripper.
I agree but Handbrake is a different animal. I know Handbrake uses all 16 cores and 32 threads of the 1950X at 100% usage when transcoding HEVC 4K and 1080P at very high settings. For some reason, the 3970X didn't offer a sizeable improvement here. I've heard Handbrake can't utilize more than 16 cores but that is what I've read. I don't have a 3970X to see for myself.
 
Handbrake isn't great for cpu. You should look at using staxrip which can take advantage of massive core counts.
I've looked at Staxrip. The problem with utilizing the GPU is a degradation in quality. I'm a quality snob and the best video quality is when left to the CPU to do the work.
 
I've looked at Staxrip. The problem with utilizing the GPU is a degradation in quality. I'm a quality snob and the best video quality is when left to the CPU to do the work.

Yea? So you should be using staxrip with a high core count cpu man. Ask tangoseal about staxrip.
 
I agree but Handbrake is a different animal. I know Handbrake uses all 16 cores and 32 threads of the 1950X at 100% usage when transcoding HEVC 4K and 1080P at very high settings. For some reason, the 3970X didn't offer a sizeable improvement here. I've heard Handbrake can't utilize more than 16 cores but that is what I've read. I don't have a 3970X to see for myself.

If Handbrake can't use more than 16 cores, it's pretty much pointless. Any 16 core CPU will only show minimal gains over the 1950X at present. That's a very narrow metric to judge those CPU's by. That's why I said there would be a massive increase between a Threadripper and a Zen 3 based CPU. I based that statement on a variety of tests I conducted myself for my review of the 3900X. I tested the 12c/24t 2920X against the 3900X when Zen 2 launched and the latter smoked the Threadripper 2920X in every test. It generally did so by a wide margin. One of the reasons for that is the superior layout of the CPU and a reduction in latency.

I mean, if Handbrake is all you care about then fine. But that's hardly a way to measure a CPU. Asking for comparisons of CPU's that have higher core counts for an application that can't utilize them makes no sense to me.
 
If Handbrake can't use more than 16 cores, it's pretty much pointless. Any 16 core CPU will only show minimal gains over the 1950X at present. That's a very narrow metric to judge those CPU's by. That's why I said there would be a massive increase between a Threadripper and a Zen 3 based CPU. I based that statement on a variety of tests I conducted myself for my review of the 3900X. I tested the 12c/24t 2920X against the 3900X when Zen 2 launched and the latter smoked the Threadripper 2920X in every test. It generally did so by a wide margin. One of the reasons for that is the superior layout of the CPU and a reduction in latency.

I mean, if Handbrake is all you care about then fine. But that's hardly a way to measure a CPU. Asking for comparisons of CPU's that have higher core counts for an application that can't utilize them makes no sense to me.
I hear you. My system is very task specific with video transcoding. I'm not a gamer and don't do 3D rendering or other graphics intensive applications. I'll wait for the new Threadrippers to come out. From what I've heard it looks like a release early next year...
 
I know handbrake uses 100% of my 3900X's 24 threads. It's a feature I use regularly. I convert 4k raw to h.265.

Also I can verify, although I don't have the numbers on me, that things were VERY noticeably faster between my old R7 1700 and the 3900X, but that should be very obvious.
 
I also would like to know. But from what we can see with userbenchmark.com we can expect 30% improvement.
 
Back
Top