Can someone give me some assurance that there will be a difference?

newls1

newls1

Supreme [H]ardness
Joined
Sep 8, 2003
Messages
4,607
I'm reading on other websites that the FX55 processor is a good chip (for this much money it better) but i've read some articles that say INTEL is still the way to go for gaming. I just spent 1000.00 dollars on this chip, and now to hear this, is driving me crazy. I'm hoping that when im done building my new PC, that I'll see a nice performance increase from my previous P4 3.4GHz chip (OC'ed to 3.9) and FPS increase in games. Im sorry to keep posting questions about the FX55 Chip, but I've never had an AMD chip yet, and its hard to belive that a chip that runs @ 2.6GHz, will provide better game play, and speedier performance than my INTEL that is running @ 3.9GHz, you know what I mean :confused: .. Thanks
 
FX-55 > Any P4 in terms of gaming. The FX-55 performance is amazing, rest assured!
 
Phranq said:
FX-55 > Any P4 in terms of gaming. The FX-55 performance is amazing, rest assured!
Click to expand...

Alright, starting to feel better, thanks.
 
Why the FX owns the P4:

1) The P4 has a long pipeline - that's great for core speed, but it sucks donkey for branch mispredictions.
2) The P4 does far fewer instructions per clock cycle than the FX.
3) Married to the latest DDR2 chipsets, P4 setups have latency problems in comparison to FX.
4) The P4 still uses a front side bus and an off-die memory controller.
5) The P4 produces enough heat to negate the effects of the noreaster about to hit the east coast. This is not a problem for the FX.
6) P4's still cost far too much money for what you get.
7) P4 is not a 64-bit processor.
 
Maybe you should do the research before you buy..especially when you spend 1000 bucks.

But that is just me.

/end mean post
 
Retro Rex said:
Maybe you should do the research before you buy..especially when you spend 1000 bucks.

But that is just me.

/end mean post
Click to expand...

My thoughts exactly. Must be spending someone else's money.
 
kunsunoke said:
Why the FX owns the P4:

1) The P4 has a long pipeline - that's great for core speed, but it sucks donkey for branch mispredictions.
2) The P4 does far fewer instructions per clock cycle than the FX.
3) Married to the latest DDR2 chipsets, P4 setups have latency problems in comparison to FX.
4) The P4 still uses a front side bus and an off-die memory controller.
5) The P4 produces enough heat to negate the effects of the noreaster about to hit the east coast. This is not a problem for the FX.
6) P4's still cost far too much money for what you get.
7) P4 is not a 64-bit processor.
Click to expand...

I don't agree with some points...

2) Not entirely true, both CPUs can theoretically do ~3 instructions per clk max. Also, when you are running at far higher clockspeeds, you can afford to do far less instructions per clk. How many instructions you can do per timespan effectively, also depends a lot on what CPU the code in question was optimized for. There just isn't that much P4-optimized code around yet. The Athlon64 is not a new architecture, it is an evolution of the original Athlon, which was designed to run Pentium/Pentium Pro/PII/PIII code efficiently, so there is a lot more suitable code around for Athlon64.

3) There have been articles about this... Using the latest low-latency DDR2 memory can give lower latencies than using the lowest latency DDR memory. So there are no latency problems anymore.

7) There have been 64 bit Xeons and P4s for quite some time now, and recently they hit the mainstream with the 6xx series. So this is also not true anymore.
 
Retro Rex said:
Maybe you should do the research before you buy..especially when you spend 1000 bucks.

But that is just me.

/end mean post
Click to expand...

Quoted for truth.


however you'll probably notice the most difference when upgrading your gfx card but if you've got 1k to drop on a proc you probably have 2 6800U's in SLI or something.
 
DryFire said:
Quoted for truth.


however you'll probably notice the most difference when upgrading your gfx card but if you've got 1k to drop on a proc you probably have 2 6800U's in SLI or something.
Click to expand...

Im using only 1 6800ULTRA for now, will upgrade when I have more money to spend....
 
newls1 said:
I'm reading on other websites that the FX55 processor is a good chip (for this much money it better) but i've read some articles that say INTEL is still the way to go for gaming. I just spent 1000.00 dollars on this chip, and now to hear this, is driving me crazy.
Click to expand...

Every benchmark or review I have ever seen shows the Athlon64's dominating the gaming benchmarks. This review shows that even a Athlon64 3200+ can hold it's own against anything Intel can throw at it. Your new FX55 will give you the best gaming experience you can get right now so don't sweat it. I've yet to see one of the articles that you refer to in your first post that shows otherwise as far as gaming goes.
 
kunsunoke said:
3) Married to the latest DDR2 chipsets, P4 setups have latency problems in comparison to FX.
5) The P4 produces enough heat to negate the effects of the noreaster about to hit the east coast. This is not a problem for the FX.

7) P4 is not a 64-bit processor.
Click to expand...
just some nitpicks ;)

3- the p4's problem is not latency as much as the fact that it needs an insane amount of bandwidth.. though the increased latency of ddr2 doesn't help much, and the benchmarks between ddr400 and ddr2-533 show a mixup, though ddr2 will be much faster soon, which will be good for dual core p4's which will need ungodly throughput from the ram.. if 6-7gb/s is barely enough.. sheesh

5- the fx-55 is rated for 104w.. though most will put out a bunch less than this, it's not exactly cool (compare to the winchester at 67w ;))

7- the new p4's have 64bit instructions, have a look at 'EMT64'

all that in mind, you can be assured that your fx-55 will dominate p4's in gaming (with an exception of maybe two games where the high clock speed of the p4 can almost catch up)
 
Scali said:
... There just isn't that much P4-optimized code around yet ...
Click to expand...
Are you kidding? The P4's been out for years. Plenty of code has been optimized for it, especially games, multimedia, and workstation apps.
 
rolo said:
Are you kidding? The P4's been out for years. Plenty of code has been optimized for it, especially games, multimedia, and workstation apps.
Click to expand...

QFT.

Hyperthreading comes to mind.
 
Scali said:
I don't agree with some points...

2) Not entirely true, both CPUs can theoretically do ~3 instructions per clk max. Also, when you are running at far higher clockspeeds, you can afford to do far less instructions per clk. How many instructions you can do per timespan effectively, also depends a lot on what CPU the code in question was optimized for. There just isn't that much P4-optimized code around yet. The Athlon64 is not a new architecture, it is an evolution of the original Athlon, which was designed to run Pentium/Pentium Pro/PII/PIII code efficiently, so there is a lot more suitable code around for Athlon64.

3) There have been articles about this... Using the latest low-latency DDR2 memory can give lower latencies than using the lowest latency DDR memory. So there are no latency problems anymore.

7) There have been 64 bit Xeons and P4s for quite some time now, and recently they hit the mainstream with the 6xx series. So this is also not true anymore.
Click to expand...

Scali - Could U point to some links re: your statments on 2 & 3 please. Also you are right on 7 however we have yet to se how good ANY 64 bittidness is going to be for the main stream. I may not always agree with you but its nice to see someone firmly in the Intel corner coming here to make ppl think a little bit further :D
 
rolo said:
Are you kidding? The P4's been out for years. Plenty of code has been optimized for it, especially games, multimedia, and workstation apps.
Click to expand...

I don't think many games are optimized for P4 actually. Games generally take 3-5 years to develop. P4 hasn't been around long enough yet.
That may also (partially) explain why P4 does relatively poorly in most games.

Multimedia apps... well, as we all know, P4 is strong at audio/video encoding and decoding, because those apps are indeed optimized for P4.

I'm just saying that the P4 is not necessarily slower at tasks when it loses a benchmark, it could also be that the software was not written to suit the P4. So it's slower in practice, but in theory it could perform the task faster if the software were adapted. This may also result in the Athlon64 performing the task slower. Ideally you should have optimized paths for each major architecture, not running the same code on different architectures, because that will not display the full potential of all architectures.

Besides, how many applications have been optimized for a CPU has very little to do with how long the CPU has been out. It has more to do with the need for optimizations.
 
Frallan said:
Scali - Could U point to some links re: your statments on 2 & 3 please. Also you are right on 7 however we have yet to se how good ANY 64 bittidness is going to be for the main stream. I may not always agree with you but its nice to see someone firmly in the Intel corner coming here to make ppl think a little bit further :D
Click to expand...

2) Just grab the manuals from the manufacturers and study the architecture. You'll see that it boils down to ~3 instructions that can be retired per clk on both CPUs.

3) I can't recall which site I read the article on, but Google turned up this one: http://www.xbitlabs.com/articles/memory/display/ddr2-ddr.html
Says pretty much the same I guess.
I'm sure you can find more.

As for being in the Intel corner.... on the contrary, my main PC uses an Athlon 1800+.
I am just a programmer who's interested in the inner workings of processors (in order to write better software for them), and even though I happen to own an AMD processor, I can't really say they are better than Intels... just different. I've written all kinds of test routines, and when optimizing for P4, you can really make it fly in some cases. It's a processor of extremes... The Athlon is more moderate... reminds me a bit of the original Pentium.
 
Scali said:
2) Just grab the manuals from the manufacturers and study the architecture. You'll see that it boils down to ~3 instructions that can be retired per clk on both CPUs.
Click to expand...

That's odd. I always thought the P4 could retire only 2 instructions per clock.

I agree however, that the P4 can really fly *if* you are versed in the ways of software tweaking.
 
Scali said:
I don't think many games are optimized for P4 actually. Games generally take 3-5 years to develop. P4 hasn't been around long enough yet.
That may also (partially) explain why P4 does relatively poorly in most games.

Multimedia apps... well, as we all know, P4 is strong at audio/video encoding and decoding, because those apps are indeed optimized for P4.

I'm just saying that the P4 is not necessarily slower at tasks when it loses a benchmark, it could also be that the software was not written to suit the P4. So it's slower in practice, but in theory it could perform the task faster if the software were adapted. This may also result in the Athlon64 performing the task slower. Ideally you should have optimized paths for each major architecture, not running the same code on different architectures, because that will not display the full potential of all architectures.

Besides, how many applications have been optimized for a CPU has very little to do with how long the CPU has been out. It has more to do with the need for optimizations.
Click to expand...
first: the p4's aren't as good at gaming as the a64's purely because of how the chip is designed. one if good for some things, the other is good for others... this is something you should know ;)

second: multimedia.... is what netburst was designed for. high clock speed is good for streaming data because there isn't much variation in what has to be calculated, so assuming the clock speed and memory bandwidth is there, the netburst architecture can rip through that kind of stuff like it's nothing (though the a64 is catching up, most notably in audio encoding)

and lastly, are you implying that the p4 needs more optimizations? while that can be argued, i think more companies optimize/compile for p4 because the majority of the market has an intel system.. there's just no getting around that huge influence unless you're catering to a specific market that may have a greater bias towards amd (gamers anyone?)
 
The problem with claiming that there isn't much P4 optimized code "yet" is that ... well.. the architecture has been out for about five years. That's a very very long time in the development world.

If they haven't "done it yet" then they aren't going to. There's no point. AMD's stuck with the P6 emulation, the P-M goes back to that, and the P4... well... emulates the P6 poorly.. If Intel has half a brain in its head, it will toss the P4 like a rotten tomato and go all out on the P-M which annihilates the P-4 in every fashion. The disparity is almost pathetic, very similar to the A64/P4 comparison in real-world IPC and power.

If this optimization "can be done" why hasn't anyone actually done it yet? Does it actually work? for anything outside of a "1% of the time scenario?" would implementing this 'optimization' require a ground-up re-write of most software? Is it an assembly optimization? or can you get it done cheap by outsourced coders in VB and VC++ using standard compilers, even Intel compilers?

The optimization argument to me seems a little broken.
 
(cf)Eclipse said:
first: the p4's aren't as good at gaming as the a64's purely because of how the chip is designed. one if good for some things, the other is good for others... this is something you should know ;)
Click to expand...

That's not a very specific answer.
Can you explain in technical terms why the A64 would be designed better for games than the P4?

second: multimedia.... is what netburst was designed for. high clock speed is good for streaming data because there isn't much variation in what has to be calculated, so assuming the clock speed and memory bandwidth is there, the netburst architecture can rip through that kind of stuff like it's nothing (though the a64 is catching up, most notably in audio encoding)
Click to expand...

Multimedia is rather vague... Can't you put games under the 'multimedia' nomer aswell?
And where you say 'high clock speed', you actually mean to say 'long pipelines', I believe.

and lastly, are you implying that the p4 needs more optimizations? while that can be argued, i think more companies optimize/compile for p4 because the majority of the market has an intel system.. there's just no getting around that huge influence unless you're catering to a specific market that may have a greater bias towards amd (gamers anyone?)
Click to expand...

Obviously the P4 needs more optimizations.
And I disagree with the argument that most companies optimize/compile for P4.
First of all, there is only one compiler that is really good at generating P4 code, and that is Intel's compiler... It's not very popular among developers.
Secondly, I dare to say that most developers don't know or don't care how to rewrite their software for P4. In fact, many developers don't know a whole lot about optimizing at all. They just haven't taken the time to study how the cache of modern CPUs work, and they don't organize their data in a way that maximizes caching. In such scenarios the Athlon will always win.
 
mwarps said:
If they haven't "done it yet" then they aren't going to. There's no point. AMD's stuck with the P6 emulation, the P-M goes back to that, and the P4... well... emulates the P6 poorly.. If Intel has half a brain in its head, it will toss the P4 like a rotten tomato and go all out on the P-M which annihilates the P-4 in every fashion. The disparity is almost pathetic, very similar to the A64/P4 comparison in real-world IPC and power.
Click to expand...

You are overlooking the fact that the P-M cannot scale to clockspeeds that will make it outperform the fastest P4s available today.
You are also overlooking the fact that the P4 is facing a problem that is a physical phenomenon (leakage because of the isolation becoming too thin), not a limitation of the architecture itself.
Once the problem is under control (Intel is working on various improvements in the manufacturing processs), the P4 can scale up in clockspeed again (and consume less power, since the power is for a considerable part (about 25%) lost by leakage, not by the actual functional circuitry of the P4), and then it will once again compensate its IPC with more than enough clockspeed, as it has done before, well out of reach of P-M, and possibly Athlon64 aswell.

If this optimization "can be done" why hasn't anyone actually done it yet? Does it actually work? for anything outside of a "1% of the time scenario?" would implementing this 'optimization' require a ground-up re-write of most software? Is it an assembly optimization? or can you get it done cheap by outsourced coders in VB and VC++ using standard compilers, even Intel compilers?
Click to expand...

Sometimes a simple recompile with the Intel compiler will be enough... sometimes some routines have to be rewritten to process their data in a different order, which will give better cache-coherency... and sometimes some branches have to be traded for branch-less variations of the same code.
It's not always easy to do, but if developers are educated in this approach of writing software, it means that we can get more IPC out of our processors, and worry less about branch/cache misses... It will also mean that the P4 can then be possibly extended to an architecture with an even longer pipeline, and even higher clockspeeds.
Anyway, there's no progress in emulating a 10-year old architecture. That cannot possibly be the goal of any company.

Anyway, why does everyone act like this is new? Did you people just join in when the Athlon was introduced... or was it not as important an issue when it was Intel vs Intel rather than AMD vs Intel?
People who've been in the game longer than just a few years, will probably remember that in the days of the Pentium Pro/PII, a lot of software also ran slower than it did on a Pentium MMX of the same clockspeed. And the solution there was also to adopt a slightly different coding style. This is just the same thing all over again.
 
So imitating the P6 is not the best way to go but is the netburst archetecture the best answer we have?
 
DryFire said:
So imitating the P6 is not the best way to go but is the netburst archetecture the best answer we have?
Click to expand...

No, Itanium was a far better answer, but sadly AMD screwed that up.
Netburst is just a very interesting idea, which hasn't reached its full potential yet.
It will be interesting to see what AMD will come up with next.
But the short-term answer seems to be dual-core now.
 
Scali said:
No, Itanium was a far better answer, but sadly AMD screwed that up.
Netburst is just a very interesting idea, which hasn't reached its full potential yet.
It will be interesting to see what AMD will come up with next.
But the short-term answer seems to be dual-core now.
Click to expand...

What did AMD do to screw up Intel's Itanium?
 
robberbaron said:
This is a joke right? Itanium's 64 bit instructions are totally different from x86-64
Click to expand...

That is exactly the point!
The Itanium is a much better 64 bit processor than the Athlon64 or P4s will ever be...
But it is not that good at running legacy 32 bit x86 code. If x86-64 didn't exist, then people would automatically upgrade to the Itanium, once the limits of 32 bit were reached... Now that there is x86-64, most people will choose x86-64 over Itanium, because it runs existing apps better, even though 64 bit apps would run better on Itanium, once they would become available. And the usual chicken-and-egg problem is back... If nobody buys an Itanium, there will be no 64 bit apps on Itanium either.
So because of x86-64, Itanium is now confined to a life as a niche-processor, like all other non-x86 processors before it... while it would have been the successor to the x86 on the desktop, if Intel had its way.
 
what's to stop the rebirth of itanuim once everyone has migrated over to 64 bit?

A processor with the ability to smooth that transition seems like a very good idea.
 
DryFire said:
what's to stop the rebirth of itanuim once everyone has migrated over to 64 bit?

A processor with the ability to smooth that transition seems like a very good idea.
Click to expand...

Same reason why nobody went PowerPC or Alpha in the 32 bit x86 age. People don't want to transition at all, if possible... and if they have to, they want it to be as smooth as possible.
Itanium had to happen now. But we're going x86-64, and we'll probably be stuck with that for at least another 10 years. By then, Itanium is probably long replaced with something newer and better.
 
ia64 != x86-64
the two are completely different. the ia64 is an entirely new architecture while x86-64 is obviously just an extension to what everyone has already...

and scali: the itanium has the same issues that the nvidia's late nv30 had... while the theoretical performance is amazing, it takes a bit of specific compiling and hand tweaking to get the full performance out of the design. for companies that need insane power and don't need to run the software they write on anything else, it works.
but since re-writing code to x86-64 seems relatively simple (no, i haven't tried it yet, but i plan to soon) developers can have a much shorter down time while re-working the software, and opterons/a64's are way cheaper than the itanium is. the only roadblock now is for windows users.. microsoft is taking way longer to release the new version of windows than they should be (seriously.. it's been more than a year and a half now)
 
Scali said:
Same reason why nobody went PowerPC or Alpha in the 32 bit x86 age. People don't want to transition at all, if possible... and if they have to, they want it to be as smooth as possible.
Itanium had to happen now. But we're going x86-64, and we'll probably be stuck with that for at least another 10 years. By then, Itanium is probably long replaced with something newer and better.
Click to expand...

128 bit processing perhaps? I think the time bridge between 64 bit and 128 bit would be a lot longer than between 32 and 64.
 
(cf)Eclipse said:
and scali: the itanium has the same issues that the nvidia's late nv30 had... while the theoretical performance is amazing, it takes a bit of specific compiling and hand tweaking to get the full performance out of the design. for companies that need insane power and don't need to run the software they write on anything else, it works.
but since re-writing code to x86-64 seems relatively simple (no, i haven't tried it yet, but i plan to soon) developers can have a much shorter down time while re-working the software, and opterons/a64's are way cheaper than the itanium is. the only roadblock now is for windows users.. microsoft is taking way longer to release the new version of windows than they should be (seriously.. it's been more than a year and a half now)
Click to expand...

The Itanium, with all its issues, still beats the crap out of Athlon64, especially when it comes to floating point processing.

And if you code C/C++ for Windows, it's as simple to rewrite code for Itanium as it is for x86-64. Same OS, same compiler... just avoid the usual 64-bit pitfalls (VS.NET can actually warn for that, even when you are developing for 32-bit).

As for the price... that's also economy of scale ofcourse. If there is enough demand for the Itanium, the R&D expenses can be spread over a very large amount of units, making each unit quite cheap. Ultimately the Itanium would be cheaper, since it's a simpler design, and requires less transistors for the same performance level.
 
robberbaron said:
128 bit processing perhaps? I think the time bridge between 64 bit and 128 bit would be a lot longer than between 32 and 64.
Click to expand...

I think the limits of x86 will be reached long before we need 128 bit.
x86's main problem is that it's horribly inefficient... as we are getting closer and closer to the physical limits of silicon, we will need more efficient processors. Eventually something has got to give, and x86 should be it.
 
Back to the Creator of this thread... (I hate thread jacking and arguing like some people like to do about my raid 0 arrays :rolleyes: )


You Bought the Best (to date) CPU from AMD... Its going to be fast in EVERYTHING that you do... it may not be the 'fastest in every single applicaion' but you wont notice in anything.

The FX 55 is the best for games right now, hands down...
The P4 is the best for Encoding/Decoding etc etc...
HOWEVER
With what ever processor you have... the P4EE 3.7 or the FX55.... Its going to be fast and you will be happy... and you will NOT notice the difference... we are comparing the Ultra of Ultra high end CPUs...

Just make sure the rest of your components in your box are high end too or you wont get the most out of your processor...
 
You must log in or register to reply here.
Back
Top