Does hyperthreading effect ripping times?

S

Stupid_Newbie

Gawd
Joined
Dec 4, 2005
Messages
671
I know there is an advantage in encoding, but does it effect the creation time of an ISO? Or us the disk drive the choke? I'm asking because I just bought a dune smart d1, and will be ripping all my blurays and DVDs to ISO on my NAS. I will also be additionally ripping and re-encoding my kids movies to ITunes for ATV playback.

I'm just wondering if I would gain enough to warrent an upgrade to 2600k or 3770k
 
The bottleneck will be the DVD drive for ripping.

Not sure on the encoding part, depends on the encoder settings.
 
It's probably gonna be pretty even at encoding also, hyperthreading isn't another core so you won't see the added benefits of another core in reduced times.
 
canna said:
The bottleneck will be the DVD drive for ripping.

Not sure on the encoding part, depends on the encoder settings.
Click to expand...



This.

You might get some boost if you were creating an ISO on a hard drive, but ripping from a physical drive you'll notice nothing.
 
staticlag said:
It's probably gonna be pretty even at encoding also, hyperthreading isn't another core so you won't see the added benefits of another core in reduced times.
Click to expand...

I7's have a clear advantage over i5's for encoding, more so than can be explained by the 100MHz clock speed advantage, so it certainly helps even if it's not actually doubling up on the cores.
 
RamonGTP said:
I7's have a clear advantage over i5's for encoding, more so than can be explained by the 100MHz clock speed advantage, so it certainly helps even if it's not actually doubling up on the cores.
Click to expand...

I'm pretty sure it's the Extra L3 Cache that's helping the encoding, not the Hyperthreading. Why would hyperthreading help encoding? Encoding is a single "task" and hyperthreading is for "multi tasking" There is a reason encode jobs are qued and not all done at once.
 
ZodaEX said:
I'm pretty sure it's the Extra L3 Cache that's helping the encoding, not the Hyperthreading. Why would hyperthreading help encoding? Encoding is a single "task" and hyperthreading is for "multi tasking" There is a reason encode jobs are qued and not all done at once.
Click to expand...

hyper threading is for multiple threads which encoding can easily take full advantage of. Or are you saying since encoding is just a single "task" you can get the same performance from a single core over a quad? If that's what you're saying, you couldn't be more wrong.
 
staticlag said:
That benchmark is not using CPU encoding.

The guy is using a GT 440.
Click to expand...

Then how do you explain the performance improvement form HT? Disabling HT would not change his encoding times if he was using a GT440 to encode. Read it again. These are ENCODING times. The 440 was used as a decoder.

I would need a card with VP5 instead of VP4 (on my GT 440) to decode it faster, but since all of my encoding is done at veryslow and placebo it's not really money well spent.
Click to expand...

But here's another source if you still think HT doesn't do anything..

http://www.bleedinedge.com/forum/sh...ncoding-benches-with-Hyperthreading-Vs-non-HT
 
And just for a 3rd source... I just did a HT on/off test on my own machine (specs in sig)

I encoded A Beautiful Mind (DVD) to .m4p using Handbrake. AppleTV 3 profile with Constant Quality factor of 16.5

HT ON:
Average FPS: 114.9

HT OFF:
Average FPS: 95.2

20.69% Improvement.
 
RamonGTP said:
hyper threading is for multiple threads which encoding can easily take full advantage of. Or are you saying since encoding is just a single "task" you can get the same performance from a single core over a quad? If that's what you're saying, you couldn't be more wrong.
Click to expand...

The way I said that was a bit confusing. What I mean is that encoding benefits from your overall power, not the number of thread you have available. If you have an i3 CPU, enabling hyperthreading will not give you twice the encoding performance because you have twice as many threads. It just won't.

I could also compare the Pentium 4b 3.06 GHZ with two HT threads against a AM3 Sempron single core with one thread. Having two threads does not make the Pentium 4 faster at encoding whatsoever. The Sempron will always be much much faster dispite not having as many cores. It's all about overall power, not the how many threads nonsense.
 
ZodaEX said:
I could also compare the Pentium 4b 3.06 GHZ with two HT threads against a AM3 Sempron single core with one thread. Having two threads does not make the Pentium 4 faster at encoding whatsoever. The Sempron will always be much much faster dispite not having as many cores. It's all about overall power, not the how many threads nonsense.
Click to expand...

but you cant make that comparison, those are two completely different architectures, you cant get any more apples to oranges than that
 
ZodaEX said:
If you have an i3 CPU, enabling hyperthreading will not give you twice the encoding performance because you have twice as many threads. It just won't.
Click to expand...

I don't see anyone making that claim. Generally you're looking at 30% max benefit if the program is extremely multithreaded, as the data in this thread seems to backup.

I could also compare the Pentium 4b 3.06 GHZ with two HT threads against a AM3 Sempron single core with one thread. Having two threads does not make the Pentium 4 faster at encoding whatsoever. The Sempron will always be much much faster dispite not having as many cores. It's all about overall power, not the how many threads nonsense.
Click to expand...

You missed the part where you explain how comparing two complely different processors has anything to do with the performance benefit of Hyper-threading. Regardless of it's performance vs the Sempron, the P4 would itself be faster with HT enabled than disabled for most encoding purposes.
 
Hyper-threading absolutely increases encoding performance for encoders that can utilize many threads.

It works like this.

Intel_HTT.png


Each task (blue and green) individually take longer to complete than they would each individually, but combined, they both complete faster than doing one after the other.

So for doing video encoding if you break up the whole video job into 8 tasks rather than 4, the fact that you can inter-mix the needed execution units via hyper-threading (each core is now running 2 intermixed threads like in the picture and thus utilizing on average more of it's execution units per clock cycle) is where you can get the performance improvement.

It all depends on how diverse the execution units of the task are. If both threads just need to use the one same execution unit constantly, then you would get no gain with hyper-threading. But if each threads uses an entirely different execution unit, you could theoretically see a 100% increase in performance.
 
SirMaster said:
Hyper-threading absolutely increases encoding performance for encoders that can utilize many threads.
Click to expand...

This an important point; if an encoder is optimized for multithreading then it's possible and likely that you'll see a benefit from simply enabling hyperthreading (as the links in this thread show). In (much) older encoders that weren't, you would see little or no benefit from enabling HT.
 
ZodaEX said:
The way I said that was a bit confusing. What I mean is that encoding benefits from your overall power, not the number of thread you have available. If you have an i3 CPU, enabling hyperthreading will not give you twice the encoding performance because you have twice as many threads. It just won't.
Click to expand...

Who said anything about twice the encoding performance? Someone said it wouldn't help, I said it would and provided sources to prove it, including my own results.
 
If HT is so useless, why does it cost $100 more? Anandtech, Tom's, and many other places have shown that HyperThreading does have its uses when it comes to video editing/encoding.
 
RamonGTP said:
Who said anything about twice the encoding performance? Someone said it wouldn't help, I said it would and provided sources to prove it, including my own results.
Click to expand...

I said it. The only person i'm trying to clarify is myself.
 
You must log in or register to reply here.
Back
Top