Sign in with
Sign up | Sign in
Your question

i'll get shot for this...

Tags:
Last response: in CPUs
Share
April 15, 2006 8:02:43 AM

Well i've got nothing when it comes to cpus except lots of questions... which I've searched for but come up empty.

I've got (i'm like 80% sure) that I have an

Intel® Pentium® 4 Processor 640 with HT (3.2GHz, 2M, 800MHz FSB )

but when I used the interactive charts it gets beat out by a ton of amd chips especially all these 939 socket chips that I see people recommending on the graphic card forums.

but why do amds beat out intels with -- less operating hz, less l2 cache, and a slower RAM??? i know there is a reasonable answer and I realize I'm not educated in cpus at all but it just makes no sense to me.

More about : shot

April 15, 2006 8:11:36 AM

It's a question of how much work a chip can do. The Amd chips have what is called a short pipeline, which allows it to do more work per clockcycle.
April 15, 2006 1:02:29 PM

I woundn't put too much stock in those cpu benchmarks. If paired with a good graphics card, your P4 will play any game out there and it will perform very well in multimedia tasks. After you get above 60-70 FPS in a game your eyes will have a very hard time telling the difference so alot of the benchmark scores don't amount to much. I have a few benckmarks that I keep on an external hard drive that I use on new builds. This is mainly to stress test the machine and verify that it is performing up to its expectations. People desperate to squeeze a few more fps or "points" out of a benchmark are more concerned with the size of their "cyberpenis".

Oh yeah, AMD's outperform the current Intels because they are more efficient. However, in real life with a average user this will amount to a handfull of fps or fractions of a second.
Related resources
April 15, 2006 2:31:42 PM

It's not just the cpu. The chipset,ram and even the psu make a big diff in benches. best cpu in a junk board with bad ram will be held back.
April 15, 2006 2:49:22 PM

P4. Yuck. I had a 540. Cross your fingers and hope it doesn't fry.
April 15, 2006 3:00:00 PM

Many of the benchmarks are synthetic and merely give general ideas of a CPU's potential. Numerous other bencharks are done using real software providing authentic scenarios of the performance of a processor's capabilities.

At present AMD has a real lead, and one you will definitely notice, in a high performance environment such as gaming, in the bulk of these benchmarks. Intel tends to excel in certain video manipulation applications, though as AMD ramps up clock speeds this particular lead is less excessive than in the past. This could change in the near future and rumors to that effect abound but thus far, those rumors are just that, rumors.

If you're in the market for a new build, you may wish to study these performance tests and find the best processor for the types of software you actually use. This processor may be AMD. It may be Intel. You won't go wrong either way as long as you do your homework.

Both AMD and Intel do what they do extremely well. Never before has so much computing power been available to mere consumers and hobbiests. This is an awesome time to be a geek :lol: 
April 15, 2006 3:40:02 PM

Ducky how did you manage to fry your cpu, especially a P4? About the only truly "fried" cpu I've had brought to me was due to extreme overclocking and it wasn't a Pentium.
April 15, 2006 3:42:48 PM

I really don't think theres a "bad" CPU out there today. Both Intel and AMD have good products and in a system there are many other things that go together to make a computer. The "benchmarks" used today may or may not reflect real world apps but when you can present the same tasks to two different systems it gives a good referance to judge them buy.
For the consumer,(US!), its a win win situation. You really can't go "wrong" with either. I like AMD personally because I think I'm getting the most for my hard earned dollar. Look at the total package. What is it YOU want in your system and match up the features as a total package, the MB, storage, I/O's..
April 15, 2006 3:53:46 PM

Pipelines are one of the keys. Intel's P4 architecture was designed for Hz. When it was first released, it had clock speeds much higher than P3's, but comparable performance. Intel used a design with a very long pipeline for processing, which is basically a working queue. I don't know what the pipeline size is on the newest P4's, but it was about 22 steps when the P4 was released if I remember correctly. This means that it takes 22 clock cycles for an operation to get from the front of the queue to the end of the queue. This would be a huge performance loss, 22 Hz per operation, so the stages are parallelized, with 22 operations in process at any given time. Now, add branching and dependent logic, ie if A<2 do command x, else command y and you start to run into the problems with a long pipeline. The processor can't fill the pipeline with the next command until this command finishes, so parallelization is lost and you're back to your 22 Hz/command. Intel implemented a very advanced branch predictor, it was something like 98-99% accurate at prediction, but that 2%*22 lost cycles when it misses (probably less than 22 lost cycles, as there are a lot of other things going on to prevent lost cycles...) means there is a lot of nothing going on. Larger L2 caches are required to try and keep this long pipeline filled. The long pipeline makes waiting more expensive, so a cache miss means a lot more wasted cycles.

There are also some differences with memory controllers on die vs on MB. Intel can change memory architecture faster with a MB based memory controller and use newer memory tech, but communication to the MB-based mem controller is much slower than in processor mem controllers.

If you have highly optimized code that takes advantage of all of the capabilities of a P4, it will generally win in raw number crunching. If close attention is paid to keeping the pipeline full and all of your commands in cache those Ghz come shining through. However, this requires a large amount of assembly language tuning and almost no applications are written with these optimizations. Examples that are are typically scientific algorithms, like Folding@Home or similar.
April 15, 2006 4:12:41 PM

AMD is better with Floating point calculations


so ive heard
April 15, 2006 4:41:23 PM

Its all the same, both are good Amd and Intel, for example: put test between p4 650 and amd eqvivalent on tom's cpu charts and you'll get 2 fps difference, so therefore buy what you like more
April 15, 2006 4:51:21 PM

Quote:
Its all the same, both are good Amd and Intel, for example: put test between p4 650 and amd eqvivalent on tom's cpu charts and you'll get 2 fps difference, so therefore buy what you like more


Lols, Sorry For decending, But AMD can provide a CPU with lower "RAW SPEED" and at a Lower Price But In Performance and Quality AMD beats a Bleeding Edge "DELL" With A "Former Glorift" Generation DDR setup and Still Compete with the latest Pentium EE And beat it in Majority rangeing from 75%-85% of all Benchmarks you can ever compare them in.
April 15, 2006 4:57:31 PM

Put it this way, AMD are best for games while Intel are used for games but dont compete w/ Amd. Pentium is moslty used in officed for apps, Calculations and other things that business ppl do on thier computers. :twisted:
April 15, 2006 5:05:22 PM

What benchmarks does a comparable AMD beat an comparable Intel by 75-80%?
April 15, 2006 5:07:38 PM

Quote:
Put it this way, AMD are best for games while Intel are used for games but dont compete w/ Amd. Pentium is moslty used in officed for apps, Calculations and other things that business ppl do on thier computers. :twisted:


AMD sells for less and the Business Sides and the Programming side it can soar as well Because of their better Memory usage and FPU floatpoint and bandwith pushing
and in most systems and businessed efficient and Less powerconsuming wood and could result in 20%-30% energy savings and almost 40% savings in Heating costs. Loll "prescott"
April 15, 2006 5:10:31 PM

Quote:
What benchmarks does a comparable AMD beat an comparable Intel by 75-80%?


Umm. if you read correctly It says *let me dumb it down a bit more* They beat Intel benchmarks in 75%(percent) to 80%(percent) of all Benchmarks Not "beat an comparable Intel by 75-80%" Sorry If i typed too crudely. :lol: 
April 15, 2006 5:11:46 PM

Hey Mr. Vader (YO_KID) take it easy f00l :lol: 
April 15, 2006 5:13:28 PM

Quote:
Hey Mr. Vader (YO_KID) take it easy f00l :lol: 


He appears to be from Eastern Canada they are all like that.
April 15, 2006 5:29:52 PM

Quote:

Umm. if you read correctly It says *let me dumb it down a bit more* They beat Intel benchmarks in 75%(percent) to 80%(percent) of all Benchmarks Not "beat an comparable Intel by 75-80%" Sorry If i typed too crudely. :lol: 


Learn to type it corrrectly and I'll read it correctly. Thanks for the % clarification. Now I finally know what those funny little signs mean. So where are you getting all these %(percentages) from anyways. Any links or are you making them up? Next question: what is a "Glorift"? Care to dumb that one down for me?
April 15, 2006 5:38:59 PM

Quote:

Umm. if you read correctly It says *let me dumb it down a bit more* They beat Intel benchmarks in 75%(percent) to 80%(percent) of all Benchmarks Not "beat an comparable Intel by 75-80%" Sorry If i typed too crudely. :lol: 


Learn to type it corrrectly and I'll read it correctly. Thanks for the % clarification. Now I finally know what those funny little signs mean. So where are you getting all these %(percentages) from anyways. Any links or are you making them up? Next question: what is a "Glorift"? Care to dumb that one down for me?

Well i don't really need to lead you to Specific links because Tom's site has a very good Benchmark List. And "glorift" in my case meant "Overly Hyped" by Intel, and just turned out to be half-assed of everything it was supposed to be. Lols for Intel DDR2 was a faster version or DDR, And i guess AM2 socket had just realised that when it upgraded

Ofcouse i'm not making this up. Othse wise you know that there are Dozens of INTEL FANBOYS here to beat me down to a pulp :roll:
April 15, 2006 5:56:01 PM

Quote:
half-assed of everything it was supposed to be.


Kind of like your typing.

I was refering to the %(percentages) in you post concerning power consumption and heating. Any "dumbing" would be appreciated.
April 15, 2006 8:04:22 PM

for my humble opinion, I prefer AMD because current p4's do get enought hot to boil your cofee's water.

global warning is dangerous ;) 

but new conroe seems promising to be colder :o 
April 15, 2006 9:02:00 PM

Quote:
but why do amds beat out intels with -- less operating hz, less l2 cache, and a slower RAM???


In day to day operations, these benchmarks don't mean much. For real-work (rendering, database querying, web serving) the benchmarks even lean further towards AMD currently (maybe not true in a year) and matters most because if you can do an operation in 20% less time and 40% less power you've saved buckets of cash.

http://www.gamepc.com/labs/print_content.asp?id=paxvill...

I can vouch for the Maya rendering times - the Opterons are that much faster and also cheaper and also cooler. And Apache is the most popular (I'd say best) webserver out there... Xeons systems can be cheaper if you can negotiate a discount from a vendor like HP or Dell.

For home use, who cares. Buy the cheapest.
April 15, 2006 9:04:31 PM

Quote:
but new conroe seems promising to be colder :o 


I want Intel to continue that trend to force AMD to release their HE models at the same price point. Fast and cool CPUs for everybody!! If you've ever walked into a room full of Xeon CPUs, you'll understand why colder is better.
April 15, 2006 10:20:37 PM

Quote:
Well i've got nothing when it comes to cpus except lots of questions... which I've searched for but come up empty.

I've got (i'm like 80% sure) that I have an

Intel® Pentium® 4 Processor 640 with HT (3.2GHz, 2M, 800MHz FSB )

but when I used the interactive charts it gets beat out by a ton of amd chips especially all these 939 socket chips that I see people recommending on the graphic card forums.

but why do amds beat out intels with -- less operating hz, less l2 cache, and a slower RAM??? i know there is a reasonable answer and I realize I'm not educated in cpus at all but it just makes no sense to me.

AMD's have far fewer pipeline stages, therefore the Netburst frequencies are bloated and get nothing done despite the massive clocks.
Also, AMD's have a lot more L1 cache, translating into superb gaming performance. Even without this, however, their current L2 cache's should still easily outperform Intel's due to cache latency being far lower on K8 architecture.
This is also important in DDR2. The memory a chip uses makes little to no difference, provided it is not bottlenecked. DDR1 had low latency and did not create bottlenecks for AMD, thus it was not necessary (and still is not necessary) to switch. For those AM2 fans expecting 30% more performance because you bought a DDR2 set, you are SOL.
In the fields of overclocking, however, Netburst is still king, belive that or not. AMD uses 90nm SOI versus Intel's Strained Silicon, which I find had poorer overclocking yields. The fastest PC's on the planet with cascade cooling are all on Strained Silicon chips, FX series and Intel's, and the best PI times you can find are all on Netburst oriented chips/
April 16, 2006 1:01:27 AM

Quote:
AMD's have far fewer pipeline stages, therefore the Netburst frequencies are bloated and get nothing done despite the massive clocks.


Let's face it -- AMD has been designed to execute legacy code, while Netburst has been made having optimized software in mind. Unfortunately, optimized software come too little too late at least for Netburst. That is unfortunate because we would all benefit from optimized software instead of slow bloatware compiled for 386 which we still have to deal with.

As a programmer who is into code optimization, I can tell you that for highly optimized code, Netburst can still perform better than AMD.
One example is the forward projection code that we wrote -- on Pentium 630 running at stock 3GHz it takes 253 seconds, and on Venice 3000+ running at 2.5GHz with TCCD memory it takes 281 seconds to complete. It would probably take 390 seconds if it was running at default 1.8GHz with DDR400, not to mention that 630 can be overclocked as well.
April 16, 2006 2:40:20 AM

Still looks like clock for clock the AMD chip is faster.... didn't do the math, just glanced at it.
April 16, 2006 6:58:17 AM

Quote:
AMD's have far fewer pipeline stages, therefore the Netburst frequencies are bloated and get nothing done despite the massive clocks.


Let's face it -- AMD has been designed to execute legacy code, while Netburst has been made having optimized software in mind. Unfortunately, optimized software come too little too late at least for Netburst. That is unfortunate because we would all benefit from optimized software instead of slow bloatware compiled for 386 which we still have to deal with.

As a programmer who is into code optimization, I can tell you that for highly optimized code, Netburst can still perform better than AMD.
One example is the forward projection code that we wrote -- on Pentium 630 running at stock 3GHz it takes 253 seconds, and on Venice 3000+ running at 2.5GHz with TCCD memory it takes 281 seconds to complete. It would probably take 390 seconds if it was running at default 1.8GHz with DDR400, not to mention that 630 can be overclocked as well.
Yes, Intel can smoke AMD in some clock-speed oriented programs with its high frequencies. This is true and always will be. However, in any situation requiring low-latency, AMD just dominates. AMD wins most scenarios, that's the point. In AMD's defense, you should have paired it with UTT chips and a low memory multiplier ;) 
Intel learned the hard way and saw that low, efficient frequencies are king for thermal management/speed, thus we have Conroe. BTW, who do you work for and what kind of code do you optimize?
April 16, 2006 7:03:26 AM

I am eagerly awaiting for AMD to drop prices to where a Opteron 180 will cost around $450...
just a month or two more should be dropping...
April 16, 2006 7:07:09 AM

Quote:
I am eagerly awaiting for AMD to drop prices to where a Opteron 180 will cost around $450...
just a month or two more should be dropping...

Why do you want an Opteron 180? The 170's are the best of the dual-core Opteron bunch. They have the best steppings that usually hit the highest. You are getting it for overclocking, right?
April 16, 2006 7:13:18 AM

Yeah, should of said the 185.
My 175 runs stable at 2520, and I want a stable chip running 24/7 at around 2800 so I figure the best way is buy one closer to that speed and by using less of a overclock, greater chances of increased stability.
April 16, 2006 7:29:21 AM

I'm really not convinced. The only dual core Opteron's that will do 3.0 are the 170's. The rating process isn't as precise as you would expect.
For example, my Venice will do 2.9, but it seems like only 3200's are capable of doing this, despite being the same core as the 3000's, 3500's and 3700's.
Your best bet is getting it off ebay and asking for the core stepping so you can do research. AMD can switch steppings and you could be screwed. Regardless, you might be better off switching to a low-end Conroe for $450 anyway. Just remember overclocking is synonymous for "luck"
April 16, 2006 8:06:41 AM

Just FYI.... my 165 is stable at 2.7ghz all day long, I can't say the memory controller wants to run my memory at 500 which is what it is rated for, but at the same time I have only tested my memory in one system so I am not sure where the problem is coming from.
April 16, 2006 8:09:44 AM

Quote:
Just FYI.... my 165 is stable at 2.7ghz all day long, I can't say the memory controller wants to run my memory at 500 which is what it is rated for, but at the same time I have only tested my memory in one system so I am not sure where the problem is coming from.

Stepping? What memory? The memory controller I used on the Operon was the same overall as the E3 Venice, and it ran DDR520 fine.
April 16, 2006 8:26:28 AM

Quote:
As a programmer who is into code optimization, I can tell you that for highly optimized code, Netburst can still perform better than AMD.


Optimizing for the x86 architecture is a bit of joke (the compilers are so darn good these days - only need to worry about cache misses and branch mispredictions) unless you are writing SSE1/2/3 code. What's great about Conroe is that it is based on the Pentium M which is optimized for PIII code which most apps I believe have been tuned for. I've always admired the P-M - low power and could easily kicks it's older brother's butt if bumped up to the same Ghz.
April 16, 2006 9:11:50 AM

Quote:
As a programmer who is into code optimization, I can tell you that for highly optimized code, Netburst can still perform better than AMD.


Optimizing for the x86 architecture is a bit of joke (the compilers are so darn good these days - only need to worry about cache misses and branch mispredictions) unless you are writing SSE1/2/3 code. What's great about Conroe is that it is based on the Pentium M which is optimized for PIII code which most apps I believe have been tuned for. I've always admired the P-M - low power and could easily kicks it's older brother's butt if bumped up to the same Ghz.
Good luck getting a Dothan to 3.0. Regardless, P'M is actually more efficent than K8.
April 16, 2006 9:27:05 AM

I say this each system has there pros and cons. I look at them like cars. Amds with v6 and the same hp then a Intel v8 with the same hp. See big motors dont mean nothing. It the power that it uses.
April 16, 2006 9:37:33 AM

Quote:
I say this each system has there pros and cons. I look at them like cars. Amds with v6 and the same hp then a Intel v8 with the same hp. See big motors dont mean nothing. It the power that it uses.

Hah, my motherboard has "V6, V8 and V12" auto overclocking settings where it auto-overclocks. Regardless though, I don't think that necessarily equates right considering it's easier to get more horsepower out of an engine with more cylinders.
April 16, 2006 12:09:39 PM

Revision F E3..... Memory is Corsair, which may be part of the problem.... On my Asus A8N Sli Premium.... I would like to try it on a DFI because I have had better luck OCing on DFI. I am using 2x1gb modules, I have also had better luck with 2x512 modules. I think some chips have better controllers on them just by luck than others as I have seen with my experience. I have a Gig of ocz ddr 600 and another gig of corsair 550. I am going to try those out also as I have had good luck with the Ocz, once I get it back from my friend. The reason I say I had better luck on DFI is cause I have another system with a 64 3000 in it which over clocked to 2.8ghz stable on an Ultra D. But on the A8N5X it will only hit about 2.5-2.6 stable. Same memory, only difference really is the board. That is my 2 cents.
April 16, 2006 12:34:15 PM

Quote:
Ducky how did you manage to fry your cpu, especially a P4? About the only truly "fried" cpu I've had brought to me was due to extreme overclocking and it wasn't a Pentium.


The computer used to be an HP. It was housed in a small case, with only an exhaust fan and vents on the side. The CPU was running at temperatures between 50 and 60C for about five months. Then it started to shut down completely at random.

I gave it to a friend of mine, who knows more about computers, and has other components to swap in and out, and he said it was a heat issue. I bought a new case, and he replaced the CPU cooler and put on some AS5. The day I got it back, I turned it on, and within five minutes, it shut down.

So now I've spent $126 on the case and labor, and another $350 on an A64, DFI board, and Thermaltake power supply. And probably another $50 for labor.

I blame all these problems on HP's crappy design.
April 16, 2006 12:36:09 PM

Quote:
Yeah, should of said the 185.
My 175 runs stable at 2520, and I want a stable chip running 24/7 at around 2800 so I figure the best way is buy one closer to that speed and by using less of a overclock, greater chances of increased stability.


You'd buy a new chip so that you can overclock it higher? Overclocking is more addicting than crack!
April 16, 2006 3:20:33 PM

Quote:
On my Asus A8N Sli Premium
Quote:

There's your problem.
April 16, 2006 4:10:45 PM

I know, but it's the only board I could get with NF4 SLI and Passive cooling.... such a nice addition to have, quite, and cool enough.
April 16, 2006 7:59:29 PM

Quote:
I woundn't put too much stock in those cpu benchmarks. If paired with a good graphics card, your P4 will play any game out there and it will perform very well in multimedia tasks. After you get above 60-70 FPS in a game your eyes will have a very hard time telling the difference so alot of the benchmark scores don't amount to much. I have a few benckmarks that I keep on an external hard drive that I use on new builds. This is mainly to stress test the machine and verify that it is performing up to its expectations. People desperate to squeeze a few more fps or "points" out of a benchmark are more concerned with the size of their "cyberpenis".

Oh yeah, AMD's outperform the current Intels because they are more efficient. However, in real life with a average user this will amount to a handfull of fps or fractions of a second.


Ditto... you nailed it on the head.
April 16, 2006 8:01:25 PM

Quote:
I know, but it's the only board I could get with NF4 SLI and Passive cooling.... such a nice addition to have, quite, and cool enough.

well, honestly sound is something I could less about. the only reason I got a fan controller was because my room reached about 10 degrees higher and since everyone else complained :p 
April 16, 2006 8:03:21 PM

Quote:
I woundn't put too much stock in those cpu benchmarks. If paired with a good graphics card, your P4 will play any game out there and it will perform very well in multimedia tasks. After you get above 60-70 FPS in a game your eyes will have a very hard time telling the difference so alot of the benchmark scores don't amount to much. I have a few benckmarks that I keep on an external hard drive that I use on new builds. This is mainly to stress test the machine and verify that it is performing up to its expectations. People desperate to squeeze a few more fps or "points" out of a benchmark are more concerned with the size of their "cyberpenis".

Oh yeah, AMD's outperform the current Intels because they are more efficient. However, in real life with a average user this will amount to a handfull of fps or fractions of a second.


Ditto... you nailed it on the head.
That's true. However, I have noticed P4's go under acceptable levels for minimum FPS much more often than AMD's do. It's not so much average FPS that counts...it's more or less the same provided you don't dip below 40.
April 16, 2006 8:11:20 PM

How does your motherboard overclock? How is that new nvidia chipset working for you???? Interesting looking stuff, just curious about them and wondering if they overclock.
April 16, 2006 8:17:37 PM

Quote:
How does your motherboard overclock? How is that new nvidia chipset working for you???? Interesting looking stuff, just curious about them and wondering if they overclock.

It works really well. The BIOS is just as featured as my old Ultra-D, and it has a lot of voltage options too. Seems like chipset doesn't really make a difference, as the 6100 has done its job. Has a lot of overclocking-oriented features, like BIOS memtest, Windows BIOS flashing, auto overclocking, auto restart failure detection, etc.
Plus the board is cheap
April 16, 2006 8:30:38 PM

jathyr whose gonna shoot you not me maybe ak with his ak47 but not me my gun has no bullets left :lol: 
!