Nforce 3 pro benchmarks w/ 244 opteron

[Pariah]

You guys got to remember that these are not x86-64 benchmarks. That will be the gravy that makes the Hammer da' King.

Sounds like the T&L argument for video cards back in the day. It's debatable that 64bit computing will have any real benefit to the home user any time soon. Even if it does, the application selection will be so poor for the forseeable future that the point will be made moot. What reason would anyone have for spending 2x or more money on an Athlon64 platform that doesn't perform any better than today's options? Because you hope the 64bit revolution will arrive within a couple months of release? A fool and his money are soon parted...

 

[dullard]

Originally posted by: Pariah
Sounds like the T&L argument for video cards back in the day. It's debatable that 64bit computing will have any real benefit to the home user any time soon. Even if it does, the application selection will be so poor for the forseeable future that the point will be made moot. What reason would anyone have for spending 2x or more money on an Athlon64 platform that doesn't perform any better than today's options? Because you hope the 64bit revolution will arrive within a couple months of release? A fool and his money are soon parted...

Exactly what I was going to say. Sure x86-64 may be the future. But when will that be? X86-64 programs aren't available now (other than a couple of OSs but you don't buy a computer just for the OS). There also aren't many programs that have been announced with x86-64 support - so it likely won't be in the coming year. Will it be 2 years, 3 years, 10 years? How long until there is enough software to justify the purchase? Will there ever be enough software to justify the purchase based on x86-64 alone? The 940 pin Athlon 64 will be a dead end without upgrade possibility if the rumors are true. It seems like the x86-64 benchmarks will come too late for the initial Athlon 64 to be AMD's "gravy".

 

[sandorski]

Originally posted by: Pariah

Originally posted by: sandorski

Originally posted by: Pariah

New generations overtaking previous generations is the exception, not the norm.

It was the norm up until the P4. Athlon trounced the K6 which trounced the K5. The PIII I don't consider a new generation, since if not for the release of the Athlon, what is the PIII probably would have still been called a PII. The PII pounded the Pentium. The original Pentiums 60/66 had some initial deployment issues, but still trashed the 486 when it came to floating point performance.

Eventually all do "trounce" previous generations, but Realworld at the time of introduction is a whole different story.

No, they all trounced right out of the gate, go look at the benchmarks again. Considering the Opteron and Athlon 64 are the same architecture running basically the same software, it would take some amazing optimizations for the stripped down Athlon 64 to be faster than the Opteron. And if it is, AMD is going to be in a tough situation, either trying to convince consumers to spend $800 on a CPU, or trying to convince corporate America to spend twice as much on a slower Opteron.

They did not "trounce" out of the gate, they did trounce later when their features(new apps which used improved FPU's, memory/cache structure, etc) and clockspeed were realized. I'm talking realworld apps, not theoretical.

 

[Pariah]

They did not "trounce" out of the gate, they did trounce later when their features(new apps which used improved FPU's, memory/cache structure, etc) and clockspeed were realized. I'm talking realworld apps, not theoretical.

Once again I refer you to this link which you seem to be having difficulty reading, for some unknown reason: The Empire Strikes Back - Intel's Pentium II CPU

WinQuake 1.09 Timedemo2 640x480

Pentium II 300: 32.7FPS
Pentium MMX 200: 15.9

For the math impaired that's a 106% improvement in a realworld app. What exactly is your definition of trouncing if more than double the performance doesn't qualify? Notice that isn't some BS 300fps to 400fps increase that no one cares about. 15.9 - 32.7 is a very beneficial real world performance improvement from practically unplayable to very playable. I don't think we need to rehash the performance improvement going from a K6 2+ 450 to an Athlon 600MHz which was a complete rout in the Athlon's favor. As for the K5 to K6, the K5 is too old to find benchmarks online, but the K6 likely saved AMD because the K5 was so late and slow that it almost put AMD out of business. Do you really need benchmarks based on that to prove there was a dramatic increase in performance? This is such a ridiculous argument that I can't even figure out what you are trying to base yours on besides just arguing for the sake of arguing.

Believe it or not I actually found some 486 vs initial Pentium comparisons from an April 1994 online Byte issue:

April 1994 / BYTE Lab Product Report

"The overall 30 percent Windows speed advantage of Pentiums over 486s isn't the whole story: The fastest Pentiums also provide a 40 percent increase over 486s in applications such as Windows spreadsheets, which rely on strong floating-point performance. "

"The Pentium, with its two integer pipelines, advanced branch-prediction hardware, and sophisticated cache design, doubles the performance of the 486DX2-66 for integer operations, while its phenomenal FPU outdoes a 486's FPU by a factor of 4. More important, the Pentium reached this performance level without sacrificing compatibility with its immense software base."

 

[GullyFoyle]

If I am reading the first page correctly, they ran the Opteron with 2 gig of memory, and the other CPU's with 512Meg?
Shouldn't that make the Opteron results faster than if it only had 512Meg also?
And it was still beat by the P4 in many benchmarks?

Correct me if I am wrong.

 

[RagingGuardian]

I'll wait to see where the Athlon64 performance heads in the future before I try to judge it. I'm sure alot of us remember when the original P4 was beaten by the P3 when running at about the same clock speeds. Hopefully AMD can get the Athlon64 to scale well into the future.

 

[Actaeon]

Me wants!

I was very skeptical of Athlon64/Opteron. I figured I was going to go Intel for sure "next round", with Prescott and everything. But wow, these benchmarks impress me, if they can get decent yields, and a good overclocker, I'm all over it.

 

[cmdrdredd]

what I don't get is...they have to overclock in almost every worthwile benchmark (games and 3dmark which simulates gaming engines) to get a decent score close to the P4. Plus every benchmark that means crap they have some little blurb about what it exactly tests. The P4 winning benchmarks get no such info.

seriously...clock speed or no the P4 still holds well, plus the P4 is out now at a reasonable price to boot. Not to mention overclocking that 3.2 would yield amazing results.

Need I also mention that the consumer base of the PC world will see 4Ghz Intel and 2Ghz AMD and say 4>2 so Intel is better. Maybe not the whole facts but the average Joe Schmoe will not know. Plus Intel already has a majority of the OEM (Dell, HP, Gateway, many custom shops).

 

[sandorski]

Originally posted by: Pariah

They did not "trounce" out of the gate, they did trounce later when their features(new apps which used improved FPU's, memory/cache structure, etc) and clockspeed were realized. I'm talking realworld apps, not theoretical.

Once again I refer you to this link which you seem to be having difficulty reading, for some unknown reason: The Empire Strikes Back - Intel's Pentium II CPU

WinQuake 1.09 Timedemo2 640x480

Pentium II 300: 32.7FPS
Pentium MMX 200: 15.9

For the math impaired that's a 106% improvement in a realworld app. What exactly is your definition of trouncing if more than double the performance doesn't qualify? Notice that isn't some BS 300fps to 400fps increase that no one cares about. 15.9 - 32.7 is a very beneficial real world performance improvement from practically unplayable to very playable. I don't think we need to rehash the performance improvement going from a K6 2+ 450 to an Athlon 600MHz which was a complete rout in the Athlon's favor. As for the K5 to K6, the K5 is too old to find benchmarks online, but the K6 likely saved AMD because the K5 was so late and slow that it almost put AMD out of business. Do you really need benchmarks based on that to prove there was a dramatic increase in performance? This is such a ridiculous argument that I can't even figure out what you are trying to base yours on besides just arguing for the sake of arguing.

Believe it or not I actually found some 486 vs initial Pentium comparisons from an April 1994 online Byte issue:

April 1994 / BYTE Lab Product Report

"The overall 30 percent Windows speed advantage of Pentiums over 486s isn't the whole story: The fastest Pentiums also provide a 40 percent increase over 486s in applications such as Windows spreadsheets, which rely on strong floating-point performance. "

"The Pentium, with its two integer pipelines, advanced branch-prediction hardware, and sophisticated cache design, doubles the performance of the 486DX2-66 for integer operations, while its phenomenal FPU outdoes a 486's FPU by a factor of 4. More important, the Pentium reached this performance level without sacrificing compatibility with its immense software base."

Compare a Pentium 233 MX to a P2 233. The 486 maintained the speed advantage for some time over Pentium, due to superior clock speeds.

 

[Lyfer]

Looks promising, I hope they're able to bring the prices down.

 

[GullyFoyle]

seriously...clock speed or no the P4 still holds well, plus the P4 is out now at a reasonable price to boot. Not to mention overclocking that 3.2 would yield amazing results.

I don't know if $700 for the P4 3.2 is all that reasonable (newegg price), but it's still cheaper than the Opteron ($826 for the 244).

Maybe I'm just a cheap-azz (I'm still running an XP1700, now available for ~$40)

 

[GoodRevrnd]

Originally posted by: Pariah
Nothing too exciting. Gone are the days of new generations that beat the incumbents into submission. Who cares if it can somtimes beat/sometimes loses to a CPU 1.5GHz faster? The fact is, it is running 1.5GHz slower and isn't going to make that clock speed up next week. Wake me up when the Opteron/Athlon 64 "Northwood" shows up.

Ditto. Nothing utterly disappointing, but nothing exciting either. At least I hope it's a good springboard for AMD to make something of the chipset.

 

[Pariah]

Compare a Pentium 233 MX to a P2 233. The 486 maintained the speed advantage for some time over Pentium, due to superior clock speeds.

Why would anyone compare the slowest available of one to the fastest available of another? What sense does that make? That and the fact that the Pentium 233 MMX was released after the initial Pentium II's (233,266,300) makes it even more of a worthless comparison. Also, the 486 never at any point had a clock speed advantage from the 1st day the Pentium was released:

History of the Microprocessor

So basically you have proven you have know idea what you are talking about, and have provided no evidence to back up anything you have claimed making any further discussion about this not worth the effort.

 

[sandorski]

Originally posted by: Pariah

Compare a Pentium 233 MX to a P2 233. The 486 maintained the speed advantage for some time over Pentium, due to superior clock speeds.

Why would anyone compare the slowest available of one to the fastest available of another? What sense does that make? That and the fact that the Pentium 233 MMX was released after the initial Pentium II's (233,266,300) makes it even more of a worthless comparison. Also, the 486 never at any point had a clock speed advantage from the 1st day the Pentium was released:

History of the Microprocessor

So basically you have proven you have know idea what you are talking about, and have provided no evidence to back up anything you have claimed making any further discussion about this not worth the effort.

We're talking generations here, we all are aware that each new generation overcame the previous, mostly after a period of time.

The 486 was faster than the Pentium for a period after the Pentium was released, reaching speeds up to 133mhz.

The Pentium topped out at 233mhz, at the same 233mhz the P2 was nearly identical in performance. Sure, P3 came in at 300mhz and thusly was faster than a Pentium 200/233), but as a Generation al difference you must compare the 2 at the same available speed, 233mhz.

On the face of it, I can see how this comparison may seem whack and inconsistent, giving the 486 the advantage of it's full mhz while ignoring the same kind of allowance for the P2/Pentium comparison. Remember, the discussion is about how a New generation compares to the Old, not whether the New has the capability to beat the Old, the New always beats the Old given time.

 

[grant2]

What's the be impressed by? Opteron is "in the ballpark"

 

[Corporate Thug]

so is there any way to OC an opteron???

 

[Megatomic]

Originally posted by: dullard

Originally posted by: Pariah
Sounds like the T&L argument for video cards back in the day. It's debatable that 64bit computing will have any real benefit to the home user any time soon. Even if it does, the application selection will be so poor for the forseeable future that the point will be made moot. What reason would anyone have for spending 2x or more money on an Athlon64 platform that doesn't perform any better than today's options? Because you hope the 64bit revolution will arrive within a couple months of release? A fool and his money are soon parted...

Exactly what I was going to say. Sure x86-64 may be the future. But when will that be? X86-64 programs aren't available now (other than a couple of OSs but you don't buy a computer just for the OS). There also aren't many programs that have been announced with x86-64 support - so it likely won't be in the coming year. Will it be 2 years, 3 years, 10 years? How long until there is enough software to justify the purchase? Will there ever be enough software to justify the purchase based on x86-64 alone? The 940 pin Athlon 64 will be a dead end without upgrade possibility if the rumors are true. It seems like the x86-64 benchmarks will come too late for the initial Athlon 64 to be AMD's "gravy".

A 64bit version of UT2003 is ready and waiting for WinXP 64. More apps will follow soon enough. Besides, the Athlon64 on WinXP 64 will run 32bit apps natively. A lack of 64bit software right now is a matter of no consequence.

 

[Megatomic]

Originally posted by: Pariah

Compare a Pentium 233 MX to a P2 233. The 486 maintained the speed advantage for some time over Pentium, due to superior clock speeds.

Why would anyone compare the slowest available of one to the fastest available of another? What sense does that make? That and the fact that the Pentium 233 MMX was released after the initial Pentium II's (233,266,300) makes it even more of a worthless comparison. Also, the 486 never at any point had a clock speed advantage from the 1st day the Pentium was released:

History of the Microprocessor

So basically you have proven you have know idea what you are talking about, and have provided no evidence to back up anything you have claimed making any further discussion about this not worth the effort.

What's wrong with a clock for clock comparison? That's the only real way to compare two processors IMO.

 

[dexvx]

Originally posted by: Megatomic

Originally posted by: Pariah

Compare a Pentium 233 MX to a P2 233. The 486 maintained the speed advantage for some time over Pentium, due to superior clock speeds.

Why would anyone compare the slowest available of one to the fastest available of another? What sense does that make? That and the fact that the Pentium 233 MMX was released after the initial Pentium II's (233,266,300) makes it even more of a worthless comparison. Also, the 486 never at any point had a clock speed advantage from the 1st day the Pentium was released:

History of the Microprocessor

So basically you have proven you have know idea what you are talking about, and have provided no evidence to back up anything you have claimed making any further discussion about this not worth the effort.

What's wrong with a clock for clock comparison? That's the only real way to compare two processors IMO.

No, you compare whats given. AXP may be stuck on a lovely 2.2Ghz with high IPC, but that doesnt matter, because the P4-C can scale to 3.2Ghz. Since the fastest processor at the Pentium II launch was the 200MMX and the K6-233, you compare it to those things.

If you compare clock for clock, the Itanium II would clean house. Comparing things clock for clock is irrelevant and out of context most of the times.

 

[Megatomic]

Originally posted by: dexvx
No, you compare whats given. AXP may be stuck on a lovely 2.2Ghz with high IPC, but that doesnt matter, because the P4-C can scale to 3.2Ghz. Since the fastest processor at the Pentium II launch was the 200MMX and the K6-233, you compare it to those things.

If you compare clock for clock, the Itanium II would clean house. Comparing things clock for clock is irrelevant and out of context most of the times.

But that doesn't really tell you anything about the real worth of the architecture. I guess this is why I favor IPC over CPI, quality over quantity.

 

[Sunner]

Originally posted by: Megatomic

Originally posted by: dexvx
No, you compare whats given. AXP may be stuck on a lovely 2.2Ghz with high IPC, but that doesnt matter, because the P4-C can scale to 3.2Ghz. Since the fastest processor at the Pentium II launch was the 200MMX and the K6-233, you compare it to those things.

If you compare clock for clock, the Itanium II would clean house. Comparing things clock for clock is irrelevant and out of context most of the times.

But that doesn't really tell you anything about the real worth of the architecture. I guess this is why I favor IPC over CPI, quality over quantity.

Why exactly would you care about IPC rather than the real world end result?

Granted, if the P4 generated 200 watts of heat in order to achieve 3.2 GHz, then it would be another matter, but that's not the case, so who cares?

CPU A scales to 500 MHz with an amazing IPC of 10 and 75 watts of power dissipation
CPU B scales to 10 GHz with a "lowly" IPC of 1, and 80 watts of power dissipation

Would you still prefer CPU A simply cause of it's outstanding IPC?
To me, CPU B would obviously be the superior CPU, seeing as it will be twice as fast as CPU A in the end, given everything else like software compatibility, etc are the same.

As for Opteron vs Athlon64 vs AthlonXP vs P4 vs The World&dog vs Mr.T, why don't we all wait until we see what becomes of Athlon 64?
Opteron is meant to compete in the server/high end workstation market, hence clockspeed isn't the most important aspect, and Athlon 64 will likely always have a clockspeed advantage over Opteron, just like the P4 has over the Xeon.

 

[mchammer187]

Originally posted by: Megatomic

Originally posted by: dexvx
No, you compare whats given. AXP may be stuck on a lovely 2.2Ghz with high IPC, but that doesnt matter, because the P4-C can scale to 3.2Ghz. Since the fastest processor at the Pentium II launch was the 200MMX and the K6-233, you compare it to those things.

If you compare clock for clock, the Itanium II would clean house. Comparing things clock for clock is irrelevant and out of context most of the times.

But that doesn't really tell you anything about the real worth of the architecture. I guess this is why I favor IPC over CPI, quality over quantity.

why does it matter
i would go so far as to say a Tbird has more IPC than a Northwood P4

so Tbird is superior

i guess a G4 is better than AMD and Intel's best x86 chips because it has a higher ipc

lets say you have a proc that can do 100 IPC but can scale to 5 Mhz

but you have another that can do 20 IPC but can scale to 200 Mhz

you have to consider how far an architure will scale and its IPC

that is why a P4 is superior to AXPs

 

[Megatomic]

See you're both still looking at the 3.2GHz vs. 2.2GHz competition. For me I see 2.2GHz vs. 2.2GHz. I'm not denying that at the end of the day the P4 doesn't win out, that would be foolish. When I think about which CPU is designed better, I have to side with the one that gets more done per clock cycle. That's called efficiency and as an engineer efficiency is what it's all about.

That's all I'm saying.

 

[Megatomic]

Oh and for the record, I'm in love with the PAT enabled i865/P4C combos out there right now. If I had the money I'd buy one in a heartbeat. I'd love to o/c a 2.4c chip up to 3GHz.

 

[mchammer187]

Originally posted by: Megatomic
See you're both still looking at the 3.2GHz vs. 2.2GHz competition. For me I see 2.2GHz vs. 2.2GHz. I'm not denying that at the end of the day the P4 doesn't win out, that would be foolish. When I think about which CPU is designed better, I have to side with the one that gets more done per clock cycle. That's called efficiency and as an engineer efficiency is what it's all about.

That's all I'm saying.

3.2 or 2.2 doesnt have anything at all to do with efficiency

if you want to talk efficiency look at power consuption vs performance then centrino obviously is the best

the hammer IMO is worth anything if it can not scale past 2.2 Ghz IMO I would even say it is POORLY DESIGNED compared to a P4 who cares what approach it is real world performance that matters and

at the end of the day the amount it can scale and the IPC is what dictates a good design not one or the other