ECC buffered ram & SSE2???

[lilhobo]

i realise that this is important for servers, but is it important for video work as well??

i see a lot of people overclocking for video work and it seem a bit risky since you might get bad video sequence, in the middle somewhere...although it might not be discernable at first viewing...

Also whats does SSE2 and how many programs support it besides windvd and CCE-SP ?

[colordog]

It's not going to so much affect the video you encode, as much as destabilizing your system and causing your computer to crash or lockup.

SSE2 is an Intel-based Multimedia Extension (MMX/SIMD) that carries specific instruction sets related to handling video commands through the processor. Applications that use it gain a small advantage when running on a P4 compared to those that don't. A ton of other programs use those calls, including DVD2AVI.

[lilhobo]

so how would my

1. dual p3 1B Ghz Win2003 server SSE ECC PC133 SDRAM 512; compare to my

2. gaming rig of p4 2.4 800FSB SSE2 winXp sp1 non-ECC 512 Dual channel PC3200???

the server has SCSI but no AGP !!

HT is only 10% speed increase right??? dual channel might take it up a bit more.

[Ramirez]

If you're not dealing with mission critical data servers where an error correction is a must, there isn't much sense of using ECC-RAM, ECC-RAM modules are slightly slower then none ECC RAM< (by about 2% aprox) due to delayed data transaction needed for error correction, nevertheless you won't experience any crashes (on the contrarily, it will help to prevent some) (ECC RAM can only be used on compatible hardware)

Both Athlon (3Dnow) and P4 (SSE2) relays on the program to utilize those instruction and will gain an additional speed up in execution (pretty negligible considering modern processors clock speed (as colordog said.

You already asked that question awhile ago, and I sincerely don't understand why is so hard to believe that your HT gaming machine would be significantly faster then your Dual P3 server rig.

Have you ever heard about benchmarking programs? Try SiSoft Sandra as it seems the only way to convince you.

[colordog]

@lilhobo

You might want to Google-up some of the stuff that you're asking about to form a foundation.

Whether a motherboard has AGP or not has no bearing on encoding speed.

Dual channel, in my opinion, makes a huge difference (as well as the memory speed). This is primarily because of the massive amount of data that gets shuffled in order to encode 6 GB of video data.

Hyperthreading does not so much give a speed increase as it helps to prevent decrease. That is, I can have my encoder fully running, while at the same time buring CDs, browsing the web, etc., without it running like a snail. It allows more work per clock unit, but as it does not speed up clock units, it's not going to speed things up a lot, just let your computer be more usable during encoding.

In fact, some of the only people that I recommend buy HT are encoders. A lot of gamers believe that they'll double their FPS in Quake, etc., which is pretty silly. But I do love my P4 HT. Nothing like burning a CD at 50X while VDubMod is cranking away....

[lilhobo]

i tell you what! i NEVER trust benchmarking software..I WANT REAL WORLD EXPERIENCE

i only said AGP coz i had a r9800pro lying around and when i bought the dual p3 it couldnt be used....

i tell you what, if you are saying multitasking, then i would say the dual processor would win over HT.....

the reason i ask this again coz when i use vegas for video editing neither CPU even reached 50% load. and this is for dvd resolution huyfuy avi rendering

and video editing is a multi threaded process of video and uadio right?.....so HT would increase the speed

[b00zed]

This is relevant to the discussion: http://foldoc.doc.ic.ac.uk/foldoc/foldoc.cgi?multithreading. "Multithreading differs from multitasking..." etc.

Also you will find that two parallel SMP processors will not perform as well as one identical processor of twice the speed. Comparing a dual 1GHz P3 to a 2.4GHz P4 is a little more difficult in this case but Intel desktop CPUs aren't known for high SMP scalability. It's a nice way to gain performance for some tasks, but it requires the OS and software to support the use of two CPUs, unfortunately I on't know anything about Vegas so I can't comment on it.

Basically, you're comparing a dual 1GHz machine to a 2.4GHz machine, and claiming the dual 1GHz machine is faster. Irrespective of the IPC rates of the P3 and P4, this is highly unlikely, especially if your RAM is configured to service both CPUs through a single 133MHz bus, which bottlenecks them both horribly and you'd be lucky to see a performance increase of 50% (this is pure speculation, and at best based on observation of the scalability of P4 Xeons in parallel which is admittedly a chalk and cheese comparison.)

I'm sure anandtech has an article on SMP P3s somewhere.

[colordog]

@b00zed

well spoken.

@lilhobo

Benchmarking is real-world experience. Run a program on one-setup and see how long it takes, run it exactly the same on the other. Compare.

[lilhobo]

from that uk link, it seems that having two separate CPUs is a better option.

multithreading depends on the software written as well as the OS. and for it to work you have to have two tasks that are similar, running 2 vegas windows etc.

so burning at x50 and doing VDUB IS NOT multi-threading, it multitasking and HT is NOT used

[colordog]

@lilhobo

I must disagree with you. Encoding is split and multithreaded on my HT CPU - and that's why I'm able to have enough "processing" left to be able to burn at the same time. I'm really unsure what you base your conclusion on that's otherwise. I was certainly unable to do that before I swapped in the HT CPU, so I know for a fact that a difference is made. Divx is, for example, made to take advantage of multithreading, as well as VirtualDub

Additionally, the link provided doesn't suggest one being better than the other. All things even, it's better to have two CPUs, or four, or a zillion. As that's not cost effective, HT because a viable option. Esepcially considering that you're not limited to a select few motherboards for an HT CPU, as you are with dual CPU boards. (I've always been disappointed in the available selection.) Additionally, the money saved in buying a single processor and a single CPU motherboard gives me much more power to upgrade to a faster CPU/board combo in the future.

Two separate CPUs, in general, may be a better option, but probably not if they are out-classed, in the case of your dual P3s, versus a new P4. A dual system gives, at best, up to 50% more throughput. And a single, newer P4 is more than 50% faster than a P3. Regardless of clockspeed even, the P3 motherboard have too much buswidth limitations to come close to the new P4 boards. Again, this is where benchmarking is critical to separate opinion from fact.