New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
On a first glance they seem to be about the same in terms of performance.
The E5504 seems to be a bit newer model and it might offer some additional advantages, plus they should offer better stabilty since they are server grade CPUs in contrast where the i7 is mostly for home use.
both will be very slow but best bet would be the E5504
What format are you encoding? What software are you using? Does it use the CPU to encode or the GPU?
what else is it gonna use am quite sure its kimsufi which do not offer GPU's
@TarZZ92 even intel's integrated GPU inside the CPU could be used to encode H264 with the right software. Even a Raspberri Pi could encode H264 with it's GPU with the right software.
true but that's next to useless
A little of topic Didn't wanted to create a new thread :-)
@rds100
Hi I have an ati 5670 GPU.....When I do format conversion my gpu loads stay at 0-2% and cpu loads increases...Any way to use the GPU to accelerate my video conversion?
@noaman if you find software that can use your GPU for the video conversion - probably yes. But i don't know, i haven't played with these things.
@rds100
they are not free the ones who support :-(
the format is mp4 x264 and software using is handbrake and it uses CPU
Yep both servers are from kimsufi
I think ffmpeg (which handbrake uses?) mainly benefits from single core performance, which the i7 920 has more of (800 vs 1000 points i think). But I am not sure
As Tar said that is "useless" -- the hardware-based encoding on consumer GPUs aren't able to deliver useful compression ratios. They are about equivalent to x264 --superfast setting in quality and speed, and they cannot trade speed for quality. Superfast itself is useless for anything other than testing.
x264 itself has excellent support for multi-threaded encoding. If handbrake/ffmpeg cannot set the option properly, switch to something else.
I always invoke x264 directly (for encoding my own recorded video).
From what i have heard and read ,i7 is recommended for video encoding specifically.
haswell ones yes. but this i7 the OP is talking about is very very old ( 7 years old)
Usually GPUs help in decoding not encoding, that's always done by the CPU (mainly).
NUMA is somewhat a different beast though. I don't know how well x264 copes with it, but unless you're doing multiple encodes at the same time, it's something to consider.
Handbrake uses multithread encoding with x264, it uses all 8 cores on my Mac.
GPU's with their hundreds of cores don't work well for video encoding (or encoding for compression and quality). Since every "job" given to a core must wait for the completion of job before it (to know the optimal algorithm to use) jobs will spend most of their time waiting for other jobs to complete.
The i7-920 sees a significantly higher passmark for single thread performance.
from actual benchmarks (source: geekbench)
there is approx 10-20% increase with the 920.
True for old hardware like Sandy Bridge/Ivy Bridge, not really true for Haswell/Broadwell.
You can encode using GPU with gstreamer. Gstreamer supports encoding using VAAPI (Intel), OpenMAX (AMD Radeon), NVENC (nVidia, not included in current release, use git master). ffmpeg also supports NVENC encoding.
If you need to encode not into H.264, you should stick with nVidia GPUs which can handle H.265 as well. AFAIK newer Intel GPUs support VP8 encoding and Skylake would support VP9 encoding.
Yes but I think it does not actually make it 8 times faster. Only double or so. Can anyone try and confirm?
Don't have Mac but IIRC I was getting close to 4x speedup on quad-core (AMD Phenom).
It depends, of course, on your encoding settings. If you're not compressing very hard, and your source is raw yuv, you might possibly be I/O limited with 8 cores.
Generally, if you don't compress hard, there is a greater chance that opportunities for parallelism will be missed.
Assuming you're not I/O limited, only 2x speedup from 8 cores is ridiculous, there has to be something wrong with your configuration.
On a quadcore with HT it is around 4x faster, so scales linearly. Handbrake can also use my GPUs and says it does, however as the fans do not spin up i doubt this.