4.5ghz w/o HT or Less then 4.5ghz w/ HT?

[__Game_Start__]

4.5 amazing

 

[Graphicism]

It works at all levels of utilization.

So you can notice HT when the CPU isn't even near 100%?... that's even BS, or you are seriously putting intel processors down.

 

[crotale]

So you can notice HT when the CPU isn't even near 100%?... that's even BS, or you are seriously putting intel processors down.

Have you tried two P4 CPUs, one with HT and one without, for a extensive period?

I had to replace a 2.4C with a 2.4B one time for a week and i didn't need to run prime and a game to notice the difference.

 

[Graphicism]

Look, bottom line is, if HT was good, they would not only use it in benching... but would not have dropped it from there next line of CPUs. If you notice a dip in performance with an intel that doesn't have HT from one that does, then it doesn't leave much in the defence of intel.

EDIT: Also want to add I'm not a fan boy or whatever you call AMD lovers, because I've only had the system in my sig for about a month, and I'm already itching for Intel Dual Xeons.

 

[Tebore]

Look, bottom line is, if HT was good, they would not only use it in benching... but would not have dropped it from there next line of CPUs. If you notice a dip in performance with an intel that doesn't have HT from one that does, then it doesn't leave much in the defence of intel.

EDIT: Also want to add I'm not a fan boy or whatever you call AMD lovers, because I've only had the system in my sig for about a month, and I'm already itching for Intel Dual Xeons.

You definately don't have an idea of what HT does. I've used it and it makes a difference, even when you're not at 100%. The reason why Intel dropped it in their next line of CPUs is they improved the effiency of the cores, and they are going dual core. If they have HT on the new CPUs the die size would be outrageously large. You sir need to be open to what people are saying instead of being closed minded and dismissing the points so quickly.
Anyway...

That's one hell of an overclock 4.5ghz. I'd take the HT enabled "crippled" speed over the HT disabled 4.5ghz. The numbers they speak for themselves.

 

[Graphicism]

OK, rather than disagreing with you, in answer to the thread Vio1 - bench both 4.5ghz without HT and with at your most stable speed - and see which is fastest, I already know what the answer will be.

 

[Albuquerque]

So you can notice HT when the CPU isn't even near 100%?... that's even BS, or you are seriously putting intel processors down.

Don't start this again.

Until you have one of these processors and can watch WindowsXP boot times, application load times (even non-multiprocessor ones), or any number of other "everyday activies" happen on your machine, then you sir have nothing to say about how noticeable it is.

Any time WindowsXP or Windows 2000 has any more than ONE SINGLE process running (which is, you should know, ALL THE TIME) hyperthreading becomes "noticeable". You don't have to be at 100% utilization; hell you don't have to be at 30% utilization. Burning a CD while playing some Ogg Vorbis music in the background while also doing some work in DreamWeaver doesn't require much more than about 20% utilization -- but it's significantly more responsive with a P4HT versus a standard P4.

It's not about total processor utilization, it's about total number of threads you have running. It doesn't make your utilization higher or lower, it just makes it more accessible to all threads.

 

[Graphicism]

I'm sorry if I sound like a smart *** with all the wrong answers, but I do have some logical thinking behind what I am saying. A sort of multi-thread can be attained with any CPU by enabling a larger share of ram and processor to background processes and system cache, check out the graphic attached from windows XP performance options.

 

[Albuquerque]

I'm sorry if I sound like a smart *** with all the wrong answers, but I do have some logical thinking behind what I am saying.

Ok, so at least say something like "in my mind, it probably works this way" versus telling someone "this IS how it is". If you would have stated clearly that you were only hypothesizing, then I would have no argument at all. Logical thinking can often get you close to the answer, but don't assume that it IS the answer.

A sort of multi-thread can be attained with any CPU by enabling a larger share of ram and processor to background processes and system cache, check out the graphic attached from windows XP performance options.

We'll talk about ram first... The very first Disk Operating System (DOS) could keep multiple programs in memory at the same time. That has nothing to do with multitasking, or even more advanced, multithreading.

Further, the words multitasking and multithreading are two very discrete concepts. Multitasking is showing (by way of hardware support, or by way of cooperative/pre-emptive CPU timesharing/cyclesharing) many applications simultaneously. Multithreading is the hardware actually executing more than one thread simultaneously.

On a system with a single logical processor (any AMD, any Intel without HT enabled) your only ability is multitasking. Nearly all current operating systems run on a pre-emptive CPU timesharing system. Because you have only one processor, that means only one thread can be physically processed at any time. The kernel decides how much time each thread needs or deserves, and allocates a certain amount of processor time for each thread. To give the illusion of doing many things simultaneously, the kernel switches processor availability between all threads very quickly. While the OS is able to show multiple things it has in-queue as "running", the reality is that you're still single threaded.

On a system with more than one logical processor (any multi-core, multi-processor, or multithreaded/"hyperthreaded") you now have the ability to perform more than one task at the actual hardware level. Nearly all operating systems built in the last 2-3 years have the ability to enumerate and use multiple processors for true multithreaded multitasking. However, there is still CPU timesharing going on... If you have 40 processes and four processors, obviously those processors are still going to be cycling through a few threads each. However, the total kernel overhead when multithreading across more than one processor is still less than doing it on a single processor.

Multithreading-capable processors (such as Intel's Hyperthreading-enabled CPU's) give a boost whenever more than one thread needs to be running. But for the same reason, it doesn't increase speed when only a single thread is running. Hence the reason why hyperthreading being disabled / enabled rarely helps any one application perform much better... Only when you're performing multiple tasks (nothing to do with actual CPU utilization, only to do with multiple things happening at the same time) will a multithreaded processor show it's abilities.

The settings you took a picture of have nothing to do with making multithreading work any better, nor do they have much power to make a single-threaded CPU act like a multithreaded one.

 

[FLESHnBLOOD]

First let me say “GREAT OC”, I would love a VapoChill LS but they cost $1300+ here in Australia.

Now I haven’t got a super duper computer as you can see but I can give you some of my results (even though they are embarrassing!)

In PCMark04 – CPU Test

THIS IS WITH HT ENABLED

File Compression 4.6 MB/s
File Encryption 60.9 MB/s
File Decompression 38.5 MB/s
Image Processing 15.6 MPixels/s
Grammar Check 4.7 KB/s
File Decryption 92.7 MB/s
Audio Conversion 2469.3 KB/s
WMV Video Compression 55.1 FPS
DivX Video Compression 71.1 FPS

THIS IS WITH HT DISABLED

File Compression 3.4 MB/s
File Encryption 45.8 MB/s
File Decompression 32.8 MB/s
Image Processing 13.4 MPixels/s
Grammar Check 4.7 KB/s
File Decryption 92.0 MB/s
Audio Conversion 2451.0 KB/s
WMV Video Compression 47.9 FPS
DivX Video Compression 62.7 FPS

Let me also say that with HT ENABLED from Grammar Check down to Audio Conversion my CPU was ONLY at 50% and for the last two only 85% but with HT DISABLED my CPU WAS at 100% all the way.

I hope this helps a little

I personally would go with HT and the Lower Clock Speed.

PS. When I was doing this comparison after I had done the HT DISABLED test, I went straight to this post and with PCMark04, Windows Explorer, MS Word, MS Excel, SmartBarXP and Firefox all open at once without HT, my PC started to choke so I had to restart and re-enable it.

 

[Jimbob7]

I'm sorry if I sound like a smart *** with all the wrong answers, but I do have some logical thinking behind what I am saying. A sort of multi-thread can be attained with any CPU by enabling a larger share of ram and processor to background processes and system cache, check out the graphic attached from windows XP performance options.

Thats all good if you pc is a server...

 

[Foxie3a]

So you can notice HT when the CPU isn't even near 100%?... that's even BS, or you are seriously putting intel processors down.

This is hard to explain.

If my processor waited until it was at 100% to turn on HT, what good would it do? Nothing, because it can't go to 101%, so it's done once it touches 100%.

Lets say that HT is off. I now need to have a program do something or other to 500 files. It can do about 100 files every second without HT on.

Now lets turn on HT. Now it can do 125 files every second. So my time went from 5 seconds, to 4 seconds. I'd say that's definately noticeable.

Now any task that takes 5 seconds must be a big one since most things take so little time we dont even notice them. Think about how many things are going on. If each thing benefits just slightly, you are going to have a pretty nice performance boost since there are just SOO many things going on at any given time.

I have 291 threads right now, and that's just with Firefox, aim, and Thunderbird. I'd sure hate to have only one thread going through my pipeline at once.

 

[Vio1]

Here is a new benchie for you guys 4.5ghz

 

[Albuquerque]

Is that patched or no? 30 seconds isn't bad

 

[Turbo]

I'm quite sure the 30 sec is unpatched. I only run the unpatched version. My EE does SuperPI 1M in 29 secs @ ~4.4 Ghz and memory @ 5:4. My Prescott system ([email protected] memory @ 1:1 very tight timings) does 1M 36 secs.

In the longer tests 8M -> 32M the Prescott is faster than the EE even clock for clock though, I suspect because main memory access is faster on the Prescott.

Try running a 32M test on that 4.5 Ghz Prescott, and I'm sure you'll beat me by a good margin. My EE @ 4.4 does 32M in 29 min 5 secs. I suspect you may get as low as ~25 min with your Prescott @ 4.5 Ghz. That is if you can keep it stable there in a 32M test

 

[Vio1]

Its unpatched. 30s was not at 4.5ghz, but at 4.45ghz.
Here is the 32M test :

 

[Turbo]

Very nice 32M test indeed. The Prescott's memory access/management really shines through here

 

[Vio1]

Hi, i dont get this:

I got my OCZ booster from RMA, and it has made my computer unstable. Without the booster installed I figured out that 277fsb is the highest my ram can take at 2.85v (max on mobo). I installed the booster, left it at 2.9v and memtest freezes after 4 mins 20 sec. I was playing far cry for 5 mins and the computer went nuts. I took out the booster and memtest ran fine..... any ideas why the booster would crap on my oc?

 

[d]g[ts]

not a clue why, but have seen it myself. My guess is the fluctuations in power from the booster, but that could be isolated to my rig, and not yours.

 

[Vio1]

anyone have any ideas why the ocz booster would make my pc unstable?