Testing HyperThreading in TMPGenc ***updated Mulittasking test*** A must see **added test 1/10/2004**

[Turkish]

Wow Duvie, good post! Thanks!

 

[RussianSensation]

I personally Run Winamp, Norton AntiVirus, Seti@Client x2 units, icq, msn, and i'll have 3 internet explorers open (cause ill be downloading game demos or music), and Hardware Doctor for ABit IS7 and PLay Unreal 2 at 1024x768 on Radeon 8500 64mb with my 2.6@3.06 HT, 2x256 Geil 3500 UP

no slow down whatsoever

On top of that I have so far completed 1607 Units of Seti, and my average for 2 units is 3:10:04 (of course seti shows only the total time it doesnt matter if u do 2 units or one -- so u gotta remember in 24 hours your cpu did 48 hours of work so cpu time will be almost twice as much as real life time)

So that puts me at....1h 35min 2 seconds PER 1 UNIT OF SETI

So Hyperthreading is definately a boost in most cases for me since I always run a lot of programs at once.

 

[Duvie]

I neew to run some FH or Seti....Maybe this weekend...maybe SEtix2 + TV + Capture DVD + DVD playing.....and maybe the internet downloading some mp3s....



Waiting for some possible SpecViewperf 7.1, Maya, 3DSMax, and UT2003 testing to be added.....

 

[InlineFive]

Originally posted by: Duvie

Originally posted by: Jeff7181

Originally posted by: Accord99

Originally posted by: Duvie

Originally posted by: Jeff7181

I was expecting maybe 10% but never this high since Ifigured the intel marketing would have been exaggerating a bit. To think it should hav been more would have been hard to believe since is still just one physical cpu...

Considering the SETI@Home results I'd expect more... IIRC some people were running 2 instances of it and averaging about 2:45 per WU, and about 2:30 with one instance.

I have heard of the neative effect of HT enabledjust haven't seen it....I am planning on doing a FH test soon.....Obviously not optimised or needs to be optimised to really take effect...

Actually, I think Jeff is referring to the large benefit of HT for SETI. Using a single instance does 1 WU every 150min, with 2 instances, each instance processes 1 WU every 165min, or effectively 82min per WU, which is a massive gain. Most likely, the cause of this gain is due to the relatively poor coding of the SETI client, resulting in stalls that slows down the performance with 1 instance, but ironically enables opportunities to better utilize the P4's execution resources with 2 instances.

Right... It takes 15 minutes longer to do two at once as opposed to a single one.

well that is just flat amazing!!! I am going compared to what Intel shows at its own site and said in the marketing brief way back when.....I think that is almost approaching 2 physical cpu speeds and that does how the SETI program is coded poorly IMO....There is no way it should alomost do 2 as fast as 1 without 2 physical chips....Consider that the far left you have to toss out to go against the far right of negative benefit you have to toss out to come up with the average...


Also 15-30% in single apps and 40-90% in multitasking is nothing to balk at and still better then anything the amd chips are offering. How many of those test in encoding did the reviews run against the AMD64 in a single apps mode had those apps in hT enabled mode in the program... I would guess not many. As pointed out above many instances I can get a benefit in programs form HT is to run 2 instances and thus you know the reviews did not do that...

Wait, i'm confused here. Isn't HyperThreading restricted to Integer calculations? I though that one some P4s there was an issue with two heavy FPU threads going down both pipes. Or is this issue fixed? I would like to know because I am thinking about selling a lot of stuff and buying a 2.4C 1MB with a heavily overclockable motherboard.

 

[Accord99]

Originally posted by: PorBleemo

Wait, i'm confused here. Isn't HyperThreading restricted to Integer calculations? I though that one some P4s there was an issue with two heavy FPU threads going down both pipes. Or is this issue fixed? I would like to know because I am thinking about selling a lot of stuff and buying a 2.4C 1MB with a heavily overclockable motherboard.

No, HT works with any type of thread. How big of a gain that can be achieved is dependent on the optimization of an application, whether it's CPU limited, cache limited or memory limited and how it uses execution resources. Even for something that is well-optimized for the P4, like Prime95, running two threads of it still gives a few % increase. So in worst-case scenarios, two threads will take twice as long as one thread, but since you do two at a time, your overall throughput remains the same. But since not everything is so well-optimized as Prime95, there are plenty of opportunties for HT in multitasking situations. Typically, you can expect a 20% gain anytime you run more than 1 task at a time. And as clockspeed increase, gains from HT also increase as well.

Also, one current problem that limits the gain from HT is the small data cache size. Prescott's doubled L1 cache should greatly improve HT performance.

 

[eofeapr]

Nice work,,, thumbs up !!

Says the New-B to Duvie

 

[KF]

Duvie, you must have been reading some pretty poisonous sites for this to be such a big surprise. The Anandtech reviews at NDA time showed this type of good result for apps like this, and in the months that followed, I was seeing reports like this for encoding and SETI on this forum. It had me persuaded Intel had a remarkably significant new (and unexpected) angle to CPU performance, and I am definitely an AMD partisan. (However $50 CPUs are more my style!)

Don't expect AMD to adopt HT as such, and not because of licensing problems (which I don't believe exist.) I spent some serious time studying a couple of Intel pdfs about this project. The complexity of the design work isn't twice as much as a non-HT design, it is much more than that. And the number of possible errors to evade, and check out, is as the square of the number of design elements as compared to a non-HT version. That's because there are two sets of CPU states which are ongoing, and every single state of one virtual CPU could interact adversely with evey single state of the other.
The verification team had to invent new techniques of verification, because there was no way they could possible check every CPU x CPU state. I think even Intel execs must have worried if it could made to work correctly. Even one escaped error could have been a disaster. Notice that HT can be disabled, as a kind of insurance, and that HT is supposed to have been present in Intel CPUs long before Intel permitted it to be enabled. Dual independent CPUs, or Quad, are much simpler and easier to design. The advantage of HT is the amount of chip real estate. It is only a little more than a non-HT design, and Intel says it is by far the most improvement you can get for the amount of chip space.

To sum up, you just can't bolt-on HT. Maybe AMD will do a dual CPU on a chip, and use some HT-like ideas to reduce the chip real estate; otherwise it is too big a project. Only Intel could take on something like this.

People are mentioning games. I thought that OS threading was disabled in Wndows games (which do their own threading internally, I imagine), so HT could be of no use. Are there games that somehow use HT?

 

[Duvie]

Actually most of the lies and the myths were seen here, at www.doom9.net, and at amdforums.com (hard to believe, huh???)....


Also no site I ever saw conducted test like this, and plus what I was also showing was results can be had in single apps and if you remember early reviews including anandtechs many said that HT on often was detrimental and often yielded nothing. I am showing many apps can deliver in single appd mode ans as mentioned with seti and as seen in my besweet test if a program is not multithreaded running 2 instances can garner the strength out of it...

Find me some reviews from sites who have thoroughly tested HT out in ways like this....It is obviously a surprise to many ATers and they hang out here and read the same reviews....

 

[chsh1ca]

Originally posted by: Duvie
Actually most of the lies and the myths were seen here, at www.doom9.net, and at amdforums.com (hard to believe, huh???)....

How were they lies? Personally, I think that many people either misunderstood or outright ignored what was being tested. In almost all the benchmarks, they were testing a single application's performance with HT enabled/disabled. The results showed that in some applications HT slowed things down, and in others, it sped things up. I never once had an understanding from anandtech's review that HT would worsen overall performance in multitasking.
I really don't get why you are making a big deal of stating that apps now have optimization for something they didn't a year ago.

Also no site I ever saw conducted test like this, and plus what I was also showing was results can be had in single apps and if you remember early reviews including anandtechs many said that HT on often was detrimental and often yielded nothing.

Yes, and why are you ignoring the fact that those benchmarks were done ALMOST A YEAR AGO? I would have hoped to see a LOT more support for HT during this time, but evidently Intel hasn't pushed for it as hard as I'd expected.

I am showing many apps can deliver in single appd mode ans as mentioned with seti and as seen in my besweet test if a program is not multithreaded running 2 instances can garner the strength out of it...

Ummm, duh. Personally, I don't know why you think it's news that taking APP X, which is not multithreaded at all, and running two instances of it on a multiprocessor system (somewhat) will yield an overall increase in speed. To me, that's pretty obvious, and I should think it is so for many of you who frequent

Find me some reviews from sites who have thoroughly tested HT out in ways like this....

That wouldn't make much sense -- you are testing it out in ways that do not coincide with 'everyday use', which defeats the purpose of benchmarking the processor for daily use.

It is obviously a surprise to many ATers and they hang out here and read the same reviews....

I'm surprised it's a surprise.

It should also be noted that due to the way HyperThreading works, it could never perform the same as having two CPUs. It relies on the CPU's execution units being idle, and therefore available for use on the second thread. In some situations, this would provide a great increase when many of the execution units are simply sending data to other units -- like Media Encoding, which seems to be a large weighting of your tests. In other situations, it would mean a minimal or negligible performance increase (or none at all), because the execution units are rarely free. In yet more situations, say an app that is multithreaded - BUT - has a main thread that rarely frees up the execution units but also relies on secondary threads to do their own work, you will notice a DECREASE in performance over a single processor system. There are more in depth explanations of situations wherein you will see the decrease in performance, however, I don't think it's necessary to go into detail about these, since Anand's article on it covers it in the 'Pros and Cons' section.

 

[Duvie]

Ok lot to respond to but i am not going to quote it all....

First out of all those apps I ran plus ut2003 demo new specviewperf I have still yet to see 1 instance of HT on hurting performance in single app mode.....It may be not as much lies but misunderstanding or just ignorance to a product most did not even have but were still commenting on.....

Secondly who the hell are you to say this is not normal use...It is for me...It is for a couple ATers whom I know who commented in this thread. For the ppl who surf the net with multiple interent and watch TV or play a DVD and have seti running in the background they may not see much. IN these newer p4's the internet and the dvd playing probably doesn't even use 10% cpu usage....That being said my benches are realistic uses, I know, cause I do it often. It justifies my purchase to myself. Take from the info what you will.....

Third, No one eve stated that it was better then 2 physical cpus....From reviews and even the often overexaggerated marketing of Intel I was only expecting to see between 10% to 20%. I wasn't thinking I would see as much support in HT enabled programs as I did outside of rendering program which historically are more geared for dual cpu systems....

Finally I am sorry it wasn't news to you....It was news to others who I know read a lot of the reviews as well and must be testimony to the fact those reviews obviously did not state this or thorughly test it out.


I am still waiting a decent HT review in single apps and multi-tasking by any major review site we so frequently link to the lastest cpu reviews...

 

[chsh1ca]

Originally posted by: Duvie
First out of all those apps I ran plus ut2003 demo new specviewperf I have still yet to see 1 instance of HT on hurting performance in single app mode.....It may be not as much lies but misunderstanding or just ignorance to a product most did not even have but were still commenting on.....

You stated you felt mislead by Anand's review.

Actually most of the lies and the myths were seen here, at www.doom9.net, and at amdforums.com (hard to believe, huh???)....

Explain what is a lie, and what you view as ignorance in a review that is a year old and offers differing results from your testing. Personally, I think you are being more ignorant of the situation than Anand was November 2002. My original post was not meant to be a flame, btw, just pointing out counterpoints to your ideas, and stating my opinion.
You have yet to see HT being hurt in single app mode, why not download the benchmarks Anand mentions showing this decrease and see if things have changed? That would be a lot better way of going about things than calling the reviewers at all those sites liars or ignorant, at least IMO.

Secondly who the hell are you to say this is not normal use...

I didn't say it wasn't normal use, I said it isn't 'everyday use'. By and large, your 'average' computer user doesn't do all of the things you do. That comment (which you have conveniently taken out of context) was in reference to why Anandtech and other review sites didn't (and debatably shouldn't) test the same exact things you tested.

It is for me...It is for a couple ATers whom I know who commented in this thread. For the ppl who surf the net with multiple interent and watch TV or play a DVD and have seti running in the background they may not see much. IN these newer p4's the internet and the dvd playing probably doesn't even use 10% cpu usage....That being said my benches are realistic uses, I know, cause I do it often. It justifies my purchase to myself. Take from the info what you will.....

I wasn't bashing your choice of benchmarks, I was merely pointing out the probable reasons why a reviewer wouldn't use them.

Third, No one eve stated that it was better then 2 physical cpus....From reviews and even the often overexaggerated marketing of Intel I was only expecting to see between 10% to 20%. I wasn't thinking I would see as much support in HT enabled programs as I did outside of rendering program which historically are more geared for dual cpu systems....

Again, we're back to your perceptions. You say you were only expecting to see between 10% to 20%, yet applications (as I have said, and am getting tired of saying) have had an entire year to gear towards HT and make use of it. Are you saying that any review that disagrees with your results based on now-out of date software must therefore be utterly wrong? That seems to be what you are implying.

Finally I am sorry it wasn't news to you....It was news to others who I know read a lot of the reviews as well and must be testimony to the fact those reviews obviously did not state this or thorughly test it out.

They haven't done it recently. They DID state it, test it out thoroughly, etc, etc.. At least, in plain english for those interested in actually reading the articles, instead of looking at the graphs and making decisions. Again, I refer you to Anand's 3.06/HT review, specifically the "Pros and Cons" section.

I am still waiting a decent HT review in single apps and multi-tasking by any major review site we so frequently link to the lastest cpu reviews...

Decent meaning shows the same results that you got? That's not being very objective now is it?

 

[Duvie]

By the way I never said anadtech's REVIEW was the lies and myths.....It was meant as here in these boards the lies and myths surface, and in most cases are just shot out there from ppl who don't own the cpu but use it as a convenient mode of playing down the real benefits of HT in single apps and multitasking....

Specifically there is one higher up member at doom9.net forums who consistently tells ppl there is no benefit to run HT in single apps...not true....says that the best in any program that HT will give you is 5-10% max...not true.....

Those are the lies and myths I have tried to shed some light on...I have gottem many pms from ppl who did not know this info and they run similar apps and are very interested now in the p4c......I am happy to help those ppl....



I agree with you that there should be a larger percentage of apps taking advantage of HT without running 2 instances here a year later from HT launch....

I don't use the apps anandtech tested and the point was to use apps I used so I can see myself. It seems odd that I have tried so many and have not seen the negative effect. Is th possibility that the HT on the 3.06 may have been tweaked with additional steppings that have lead to the p4c??? maybe I don't know.....


I don't feel like arguing with you...If you don't like my comments just read the data and make up your own conclusions....Hell call me a liar!!! I don't care....I did this to help some and maybe it doesn't help all...

 

[RaynorWolfcastle]

realize that writing HT enabled apps means writing multi-threaded apps, which means that the basic structure of the program must be rethought in some cases. Writing a multi-threaded app is not AFAIK a simple "plug it in and recompile" kind of optimization, it may be significantly more demanding from a development point of view; and the thruth is that many times development time is more important than performance on a few high-end systems. This will hopefully change, but it is a sad reality right now.

 

[chsh1ca]

Originally posted by: Duvie
By the way I never said anadtech's REVIEW was the lies and myths.....It was meant as here in these boards the lies and myths surface, and in most cases are just shot out there from ppl who don't own the cpu but use it as a convenient mode of playing down the real benefits of HT in single apps and multitasking....

You constant use of 'review sites' was why I inferred you were discussing actual articles here. I find it odd you mentioned nothing about this in your previous posts, and continually made reference to reviews and review sites.

Specifically there is one higher up member at doom9.net forums who consistently tells ppl there is no benefit to run HT in single apps...not true....says that the best in any program that HT will give you is 5-10% max...not true.....

So you proved that statement wrong in three or four apps, and have concluded that it must therefore be wrong in all cases?

I don't use the apps anandtech tested and the point was to use apps I used so I can see myself. It seems odd that I have tried so many and have not seen the negative effect. Is th possibility that the HT on the 3.06 may have been tweaked with additional steppings that have lead to the p4c??? maybe I don't know.....

All you do is run mpeg encoder software?

I don't feel like arguing with you...If you don't like my comments just read the data and make up your own
conclusions....Hell call me a liar!!! I don't care....I did this to help some and maybe it doesn't help all...

Liking your comments has nothing to do with it. I think they are silly, obvious, and some of them downright wrong. I stated my opinion and reasons for having it using facts and some understanding of how HT works. Let us look at the facts:
1. You are surprised that hyperthreading offers such large performance advantage for encoding applications. This should be no surprise given even a cursory understanding of how HT works.
2. You have benchmarked only highly P4 optimized media-encoding or otherwise heavily-media intensive applications, and thus, have not stressed the execution units in the same manner as many normal applications would. This is due to the fact that the applications in question require MINIMAL execution unit use. In other words (spelled out in plain english) your pattern of usage fits the exact BEST CASE scenario for using a P4Cw/HT.
3. You titled this thread "Testing HyperThreading in TMPGenc *** A must see to dispell the myths of HT and the Lies". Yet you have given only benchmarks that fit the exact best usage for HyperThreading: MP capable benchmarks that do not stress the execution units. This is not an objective review, and does not dispell any myths of HT or Lies (as of yet unidentified via exact quote). The 'myths' you are talking about are evidently partially untrue, however your method for wholly 'dispelling' them is faulty.
4. After I suggested you download the benchmarks that Anand used (which would likely show you said performance issues you are looking for), you blew it off as being aside from the point ("I don't use the apps anandtech tested and the point was to use apps I used so I can see myself.").
In the grand scheme of all apps that will run on Windows, a total of five (seven, including UT2K3 and SPEC) represents a VERY small number. If you have tried more than those, please feel free to post the them too, I am curious.

I personally am surprised at the number of people who have responded to your thread in surprise. I had honestly thought certain individuals were interested about the technology and understood its uses. Just goes to show you learn something new everyday.

 

[Duvie]

I truly believe there was a rework of the HT from the 3.06ghz 33fsb chip which was supposed to be a enhanced xeon HT already....My reason for this belief is this link....

TMPGenc 2.8 garners only 3% with HT on

NOw th ereason I state this is that I was using an older version 2.5 freeware and I tested 4 diffeent modes from CQ to VBR to estimated motion to high quality and I garnered an average of 20% increase...It shold have scored mor like a 66 in the test instead of 80....


Maybe the mobos had poor implementation or support through the bios but something has changed and hence why I think these numebrs are disspellin g some of that first data mack with teh 3.06ghz that showed lackluster or negative results.....

Seems hard for me to believe all the apps I tested and I didn't see at least one negative result and at same time mush larger gains in similar apps......

I think the p4c's have a refined HT that may have been reworked with the d1 stepping.....

 

[Duvie]

OK...I have just spent an hour looking at reviews of HT after doing a hotbot.co search and the one thing that strikes me is that the majority if not all the reviews it found was realted to the 3.06ghz....On top of that they were done back when the 3.06ghz came out and in many cases were being done on 845e and pe chipsets which leads to my possibly thoery of some lack of bios support for it as well as the newer p4c chips and stepping may have a slight rework.

Many of the test are using older versions of programs that are out now so when they didn't get numbers or large numbers then they may have changed now...Just like the vrzone link above...

 

[chsh1ca]

You finally understand what I have been saying, lol!

Your benchmarks aren't meaningless, they just shouldn't be compared to the older benchmarks, due to different application versions, and different processor cores. I would like to see Anand do an article up on the P4C's HyperThreading, and how far it has come in a year, basically because of how well it shows in multitasking in your (albeit weighted) benchmarks.

 

[Jarhead]

Yeah, look at the revision history on this proggie!

http://www.tmpgenc.net/e_main.html

It makes full sense for a very poorly written program to see huge gains with HT on, due to portions of the CPU sitting around doing nothing....

 

[Jarhead]

Now, have you done the first basic logical thing and gone back and benchmarked with the same exact version of TMPGEnc that was used in the reviews? Could it be that the code for the program was highly tweaked since that time? Sometimes programmers can't fix the root source for issues and may have to do things that degrade the performance of thier application (dunno if this is the case here or not).


http://www.tmpgenc.net/e_main.html


Version : 2.521. 2003/9/26


Improved: In the project wizzard, for the setting concerning the bitrate, the file size in MB will be shown instead of the duration for "CD-R 74min (VCD/SVCD)" and for"CD-R 80min (VCD/SVCD)".
Improved: When error occurs during "Prefetch Video", the details will be shown.
Improved: When using "DirectShow Multimedia Reader", from now on only Linear PCM, 8/16 bit, 1/2 will be accepted for the audio input.
Fixed: Problem occuring when using an external MP2/MP3 encoder. For certain settings, the CRC value or the error protection could be changed .
Fixed: When the audio was re-sampled in high quality mode, if the input was monoral, errors could occur.
Fixed: During the encoding, some "Drag and drop" files could be accepted from the Explorer.

Version : 2.520. 2003/7/17


Fixed: Made minor corrections

Version : 2.513. 2003/6/12


Fixed: In the MPEG Setting dialogue under the GOP tab, even if the "Force picture type setting" was not enabled you could click the "setting" button.
Fixed:Reading an animation using DirectShow, could result in incorrect audio or a memory access violation error.

Version : 2.512. 2003/5/20


Fixed: When selecting an MPEG file in MPEG Tools, Multiplex, a message saying "does not support multiplexing a DVD Sub-Picture stream to a system stream" was displayed.
Fixed: In an environment where multi-thread was enabled, choosing "Highest Quality" as Motion Search Precision would cause the encoding to stop halfway with an error saying Memory access violation error.

Version : 2.511. 2003/4/28


Fixed: Adding too many chained files, would cause the program to close.
Fixed: On systems where Windows Media Player 9 is installed, reading MS MPEG-4 AVI files via the "DirectShow Multimedia File Reader" would cause an error to occur.
Fixed: Multiplexing MPEG-Video stream without a sequence end-code would cause the PTS time stamp of the picture included in the last GOP, to be invalid.
Fixed: Multiplexing a closed GOP MPEG-Video stream without a header would cause desynchronized video and audio.
Improved: Improved the processing of the file so that, if possible, securing and releasing the memory of a huge block is not repeated..

Version : 2.510. 2003/1/17


Fixed: In the Environmental setting window, General tab, checking Enable multiplex with... and chosing audio file other than mp1, mp2 or mp3 format would raise an error when starting to encode.
Fixed: In the MPEG setting dialog window, GOP structure tab, when Detect Scene Change and Force picture type setting were both selected, closing the dialog window and re-open it would clear the Force picture type setting checkmark.
Fixed: In the MPEG setting dialog window, System tab, chosing MPEG-1 Video-CD (non-standard) would in fact select the MPEG-1 Video-CD setting.

Version : 2.59. 2002/11/14


Corresponded to Hyper Threading Technology.
Improved the system of template by project wizard. Basically, it can be chosen by a combination of CBR/VBR and audio format.
It can be played by setting screen of all image filter except 24fps filter. It starts to play by pressing space bar in condition of making slider bar active.
"Set up forced picture type" and "Detect scene change" in MPEG setting can be used together.
Chained files can be managed as one animation. This tool can be switched ON by environmental setting.
Changed "external tool" tab in environmental setting to "audio engine" tab.
Included sampling wavelength transform engine with high quality. However, it is necessary to set up by "audio engine" tab in the environmental setting in order to use since it is set as not using in default.
Included the project which currently encodes in "remained job number" indicated in the status bar of main window by batch encode.
Modified not to have memory access violation error upon moving track bar or clicking stop button right after pressing a play button by audio processing dialog.
The latest setting is saved to the utmost by saving environmental setting or batch list into HDD upon job completion of batch encoding.
Message is indicated when audio file cannot be opened by the setting window of source range.
Audio corresponded to separating MPEG-2 file of linear PCM by MPEG tools. Multiplex of linear PCM cannot be supported.
Only elementary stream format can be output with WAV format by specifying linear PCM format to audio format in MPEG setting dialog.
In case "Interpolate YUV data from 4:1:1 to 4:4:4" is ON in environmental setting, "Specify color space conversion formula" is also handled as ON automatically.
Modified that TMPGEnc.vfp, which is VFAPI plug-in of TMPGEnc did not read TMPGEnc body setting.
Modified that the index after 4GB was broken when data rate output AVI file with high format (for example, uncompressed high resolution) by AVI output.

Version : 2.58. 2002/09/03

 

[Duvie]

Actually multi-threaded optimisations have been there since version 2.50 (quite awhile back)....I have been using 2.52 when I did the test and have not even tried the latest 2.59 version...The 22% increase in it withthe older version was quite a bonus in its self.....


I wasn't comparing to any specific review but I think on ly Toms hardware has ran the tmpgenc program in a review with the p4c.....

One of the biggest things I was trying to bring out here were several places and members in forums were perpetuating the myth that HT was only beneficial in multi-tasking and most single apps would gain nothing if not perhaps being a disadvantage. This was shown not to be true in this app as well as a few others.

 

[Jarhead]

Look really carefully.... 2.58 is a bogus version to be using, since they fixed it up after 2.58, specifically for HT, unless you are trying to directly compare to an old review.

2.58 is ancient, as well as 2.59 where they fiddled with the HT.

Version : 2.59. 2002/11/14
Corresponded to Hyper Threading Technology.

And the NEWEST version is Version : 2.521. 2003/9/26, not 2.59 from November 2002.


Progression of the versions...

Version : 2.521. 2003/9/26

Version : 2.520. 2003/7/17

Version : 2.513. 2003/6/12

Version : 2.512. 2003/5/20

Version : 2.511. 2003/4/28

Version : 2.510. 2003/1/17

Version : 2.59. 2002/11/14


Corresponded to Hyper Threading Technology.

Version : 2.58. 2002/09/03

 

[lookin4dlz]

Great info Duvie! I've been looking for a reason to move to HT & this provides it.

My next question is how to upgrade to get there...
Currently running a 845pe board w/ 1.8a @2.7 & 512MB Coarsair XMS memory. Easiest move would be to grab a used 3.06 off of eBay for <$200 - I hate reinstalling the hardware/software :disgust: If I can overclock this new processor to 3.2, how would that compare to a 2.4c - perhaps one of these inexpensive combo deals at pricewatch: Pentium 4 2.4GHz 800 with cpu, fan that are approximately $200?

Thanks for your input!

 

[Duvie]

Well make sure the i845pe mobo supports it, as not all of them do!!! some may need just a bios flash and others it just isn't going to happen....Also many of the boards cannot reliably do 200fsb let alone a few more fsb if you want to OC...

I say if you want to OC get yourself an inexpensive Abit IC7 or IS7 especially if you are looking at the 2.4c...the multiplier is low and it racks up to 270fsb just to get around 3.2ghz.....Many boards can do 250 or less very reliably but only look at them if you are lookingat 2.8ghz or more IMO....

A 3.2ghz is noticebaly faster then 2.4ghz in the apps I have ran in these test but I cannot speak for gaming which is likely gpu limited or the majority of the work is handled by the gpu and thus is not effected tremendously by cpu speed....

 

[TerryMathews]

For the record, while I will admit that 20-30% in TMPGenc is impressive and may convince me to jump ship, that it's in line with what it should be. My old 760MPX system would cut MPEG2 compression times by 80% in TMPEGenc. Very nice actually, I miss it. A pair of Tbred-Bs OCed to 2.26GHz and it could almost do high quality compression at 1:1 time. My current 2GHz single proc Athlon barely makes 2.6:1.

EDIT: My 2GHz system has a few performance problems that I haven't cleared up yet. Most revolving around a bad stick of RAM I'm using.

 

[Duvie]

Exactly cause it is 2 VIRTUAL CPUs not physical cpus.....If a person had a dual xeon setup they would be 4 virtual cpus and 2 physical cpus and I could only imagine the gain there....


But what you have to remember that 20-30 is pure bonus...I didn't pay for 2 cpus and I didn't have to pay fro much more expensive dual cpu motherboards or more expensive version of winxp pro....

The fact is the Barton 3200+ I tested wasn't close and I bet the athlon 64 even at 3400+ or the FX51 would not beat it....