Opteron or Athlon 64

[DannyBoy]

This is much more of a processor thread.....

....and i can see it turning into an AMD vs Intel thread....

 

[AtomicDude512]

Anyone who knows speeds?

 

[DannyBoy]

Originally posted by: AtomicDude512
Anyone who knows speeds?

Nope but I would be interested to know as well.

Perhaps the poster could EDIT the title with the question anyone know the speeds?

Or someone could create a new thread that would draw more attention?

 

[DannyBoy]

Ok im v.interested to know so ive done this thread

Dan

 

[dexvx]

Originally posted by: Snoop

If you knew the mhz of both the "240" and "242", you would be able to come to some sort of conclusion as to how much faster the "242" is...

Winz, what if AMD doubled the cache, and 'borrowed' Hyperthreading, and called it the Athlon Quarterhorse, which ran 200 mhz slower than the barton, but dominated it in the majority of benchmarks. Would MHz then be an accurate guage of performance or would an arbitrary number increase be better?????

Then that would be a different processor. He's talking about comparing CPU speeds of like processors. Your extreme example is Willamette VS HT Northwoods. The irony of this statement is that a lot of people who say Mhz is a myth, also refer to how poor the P4 performs (even nowadays) based on the first batch of willamette benchmarks.

I dont see why you shuldnt used Mhz. Higher end processors such as Sun, HP, Alpha, SGI, all use Mhz. None of the companies bothered to use a PR rating system. Centrino laptops @ 1.6Ghz match or beat 2.5Ghz P4's, but do you see intel with a PR system?

 

[DannyBoy]

Originally posted by: dexvx

Originally posted by: Snoop

If you knew the mhz of both the "240" and "242", you would be able to come to some sort of conclusion as to how much faster the "242" is...

Winz, what if AMD doubled the cache, and 'borrowed' Hyperthreading, and called it the Athlon Quarterhorse, which ran 200 mhz slower than the barton, but dominated it in the majority of benchmarks. Would MHz then be an accurate guage of performance or would an arbitrary number increase be better?????

Then that would be a different processor. He's talking about comparing CPU speeds of like processors. Your extreme example is Willamette VS HT Northwoods. The irony of this statement is that a lot of people who say Mhz is a myth, also refer to how poor the P4 performs (even nowadays) based on the first batch of willamette benchmarks.

I dont see why you shuldnt used Mhz. Higher end processors such as Sun, HP, Alpha, SGI, all use Mhz. None of the companies bothered to use a PR rating system. Centrino laptops @ 1.6Ghz match or beat 2.5Ghz P4's, but do you see intel with a PR system?

No but I would still like to be able to compare the speeds of the new opteron's

 

[jhu]

I dont see why you shuldnt used Mhz. Higher end processors such as Sun, HP, Alpha, SGI, all use Mhz. None of the companies bothered to use a PR rating system. Centrino laptops @ 1.6Ghz match or beat 2.5Ghz P4's, but do you see intel with a PR system?

those companies emphasize the spec benchmarks more than mhz.

 

[Wingznut]

Originally posted by: Snoop

If you knew the mhz of both the "240" and "242", you would be able to come to some sort of conclusion as to how much faster the "242" is...

Winz, what if AMD doubled the cache, and 'borrowed' Hyperthreading, and called it the Athlon Quarterhorse, which ran 200 mhz slower than the barton, but dominated it in the majority of benchmarks. Would MHz then be an accurate guage of performance or would an arbitrary number increase be better?????

But they didn't. There are no architectural differences between the 240 and 242.

And you are using this hypothetical example as if I had said that mhz was the end-all, be-all form of measuring cpu's performance. And actually, I said very differently in my post that you chose to quote only part of. So, I'll say it again...

Mhz does not tell the whole story. But it is a significant part of said story. Much more significant than an arbitrary number.

Originally posted by: jhu
I dont see why you shuldnt used Mhz. Higher end processors such as Sun, HP, Alpha, SGI, all use Mhz. None of the companies bothered to use a PR rating system. Centrino laptops @ 1.6Ghz match or beat 2.5Ghz P4's, but do you see intel with a PR system?

those companies emphasize the spec benchmarks more than mhz.

Very true. But they do still use mhz as part of the equation.

 

[CBone]

Originally posted by: Wingznut

Originally posted by: Snoop

If you knew the mhz of both the "240" and "242", you would be able to come to some sort of conclusion as to how much faster the "242" is...

Winz, what if AMD doubled the cache, and 'borrowed' Hyperthreading, and called it the Athlon Quarterhorse, which ran 200 mhz slower than the barton, but dominated it in the majority of benchmarks. Would MHz then be an accurate guage of performance or would an arbitrary number increase be better?????

But they didn't. There are no architectural differences between the 240 and 242.

Where did you find this out? I thought that they varied within the same group by frequency, HT bandwidth and/or cache size. Wouldn't that make them a different architecture much like the northwood and willamette are different architectures? I would figure that the 240 = old 400FSB 256K willamette and 242 = 533FSB 512K northwood. If my reasoning is flawed, that is a huge price gap for there to be no architecture differences between the chips.

 

[Markfw]

I for one am just going to wait for April 22nd so I don't go crazy guessing at everything. I guess if you have nothing better to do that argue the unknown, then go for it.

I am really curious however.......

 

[Mrburns2007]

Before a pro buys a Opteron he will research the 240, 242 and 244 and see how the real benchmarks are for his need.


The Opteron is aimed at the "pro" not home users so hopeful there suppose to be a little smarter then average joe who gets caught up in the mhz only spec.

 

[SUOrangeman]

In sum, I understand that the first digit of the model number addresses the SMP nature of the chip. "2" indicates that the chip will work in a dual-processor configuration.

No one has figured out the middle digit, and assume that it will always be 4. If I had to guess, I's say it was related to the actualy chip speed, presented as a multiple of 500Mhz. 4 x 500Mhz = 2GHz.

I have an inkling that the last digit was something to do with the size of the on-chip cache. 0 may be 256KB or 512KB, 2 = 2MB, and 4 = 4MB. My only justification is the tremendous price increase from the 240 model to that of the 242. Cache is still very expense ... otherwise, we'd do everything in cache!

Oh well, enough guessing for now.

-SUO

 

[Armitage]

FWIW, the quote I got from Appro was for a 1.4GHz Opteron. They used the 2600+ rating. I suspect this corrosponds to the "Model 240" but my quote didn't have that notation. On the phone they mentioned that a higher speed chip was being released at the same time, but with a significant price premium. Don't recall if that was 1.6 or 1.8 GHz, but I suspect that is the "Model 242". Both have 1MB L2 cache

But, as others have said, if you're going to make a purchase, you do it on the basis of benchmarks, not MHz, or names generated by the PR department.

 

[PrinceXizor]

This is eerily similar to another thread when the naming convention for the Opteron chips was released.

It was mentioned in this previous thread that current server chips from other manufacturers use a similar system consisting of the # of multiple processors, core reference and placement performance wise within a family of processors.

P-X

 

[Wingznut]

c|net news.com link

The Opteron 240 will run at 1.4GHz and cost around $340 in volume quantities, said sources, while the Opteron 242 and 244 will run at, respectively, 1.6GHz and 1.8GHz and cost around $800 and $900.

 

[ElFenix]

Originally posted by: Wingznut
c|net news.com link

The Opteron 240 will run at 1.4GHz and cost around $340 in volume quantities, said sources, while the Opteron 242 and 244 will run at, respectively, 1.6GHz and 1.8GHz and cost around $800 and $900.

looks like they're all coming with 1 meg cache.

the mhz is a bit worrisome, i know they were shooting for 2GHz for launch and it doesn't look like they've hit it.

either that or the processor is performing better than expected. guess we'll have to wait until tuesday.

 

[Wingznut]

Originally posted by: ElFenix
looks like they're all coming with 1 meg cache.

Think so? I figured the 1.4ghz would be of the 256k flavor.

And the low mhz isn't really surprising... From what I've heard, SOI has been a real bitch for them, and IBM as well.

 

[LeeTJ]

Originally posted by: LikeLinus

Originally posted by: Snoop

Great. That's AMD for ya! An even more confusing sliding scale for performance based on an imaginary starting point

It could easily be argued that basing performance on MHz is equally as vague. Simply look at how the P4 has evolved; with each new stepping, performance did not scale directly to MHz, so at least this scaling system will give values relative to its predecessors.

At least theirs is a "real" number and scale that has been used for many years and many companies. AMD's scale is self sufficent and only to their needs. It's not to help the customer but to help themselves look better. Plus having 3 different numbers scale for 1 chip is confusing. Pentium as P4 and P4 Mobile or Xeon. It doesnt have P4 100 p4 200 p4 300. Ontop of that it's then done by scale of performance (i.e. p4 140 p4 240 p4 340). It's much easier just to name the chips different than to have like 20 different number names that are confusing to the end customer.

I love AMD and their chips, but the names they pick and numbering isn't helpful.

Don't you mean SELF SERVING??

 

[LeeTJ]

Originally posted by: Wingznut

Originally posted by: Snoop

Duh, yeah of course you could always check the system specs. But it'd be like GM coming up with a new name for the V6. Of course there are V6 out there that outperform V8's and It doesn't really depend on what the "number" is after the v, but it gives the buyer a good sense of reference to the assumed power that the car can output. So if GM started calling their engines 140, 242, 340 and making the people LOOK for the actual displacement, how annoying and stupid would that be?

In what way is MHz a good sense of reference? MHz mean nothing to the performance of a cpu.

So, you would have no idea how a P4-2.66ghz would compare to a P4-2.8ghz?

Using mhz as a point of reference between two completely different processes doesn't work so well, but using it amongst the same family is very relevant.

MUCH more relevant than adding a 2 to the model number.

really?? and i can infer from this that a 3.06 P4 is superior to a 1 ghz IA2??

MHZ are overrated.

 

[Wingznut]

Lee, I think you missed the part where I said "Using mhz as a point of reference between two completely different processes doesn't work so well, but using it amongst the same family is very relevant." The Itanium and P4 are very different processors, and comparing them (mhz to mhz) is irrelevant.

But if you want to compare a 1.0ghz Itanium 2 with a 900mhz Itanium 2, you can easily draw some sort of conclusion as to the relative performance.

Mhz being "overrated" depends on your perspective. If you understand the differences between different architectures, then you can put the correct (relative) emphasis on the clockspeed. But if you think that mhz is the primary measurement of all cpus' performance, then yes, you are definitely overrating it's validity in the comparison.

 

[Jeff7181]

AMD's Pricing

And for those of you confused with AMD's Opteron rating system... remember that the Opteron is a server level CPU, and they could care less if the average home user can understand the system or not. I'm pretty sure with the Athlon-64 they will use a similar rating system to the one used now. Maybe they'll change it a little bit, but I'm sure it will be fairly easy to understand so they don't drive away customers with confusion.