Reminds me of.........

Toggle sidebar Toggle sidebar
T

thelastguardian

Distinguished
May 7, 2002
44
0
18,530
Right now the state of Hammer kind of reminds me of before P4 comes out, everyone and amd say how great the 64 Hammer cpu are really reminds me of before P4 comes out how everyone was saying how great the P4's Netburst are. I am kind of worry about amd's Hammer wouldnt turn out as good as it sounds. If thats the case then Amd is screw. You guys dont have to agree with me but thats what i feel anyway, you guys are welcome to drop some insights.
 
Some hints why it won't be a P4 flop-like:
It's a K7 with so much improvements. Where can it be bad? Nowhere. K7 is an awesome piece of performance, now K8 upgrades it drastically.
No huge pipeline also contributes to a good IPC, even if it has 2 more stages.
More cache, SSE 2, strong FPU makes up for what P4s kept doing, by using the SSE 2 to hide their low FPUs, while Hammer has plenty, AND SSE 2 as seperate, so the performance improves better.
0.13m+SOI+Heat Spreaders contribute to even higher clockspeeds without any CPU problems, HS contributes also to prevent CPU meltdowns, I think.
On-die mem controller makes up for almost 20% higher performance, that's something which they cannot just cancel like on P4's stripped components.

There are many more improvements in Hammer, which is far from what Willamette had done. Intel had hyped it having so much, but it turned out so huge, they stripped it to hell. K8 is already demonstrated, it is smaller than a 0.18m AthlonXP, possibly smaller than Northwoods too, I forget. This means it won't have huge costs but with SOI it might level back up, AND it is packed with features. I doubt AMD would trash their current steppings, and remake the core huge, and trash all the new features!

--
Luke, I am your father...but due to a bacon-slicing accident, your mother... :lol:
 
That post reminds me of the kind of thing that people were saying about the P4, too.

I completely agree, the Hammer is way overhyped. Does that mean it'll suck? Not at all, but look how excited everyone was about the nForce, and how it really is. Similar thing.

<font color=blue>Hi mom!</font color=blue>
 
K7 is an awesome piece of performance, now K8 upgrades it drastically.
Click to expand...
You're forgetting that AMD has publicly stated that Clawhammer is expected to be approximately 20-25% faster than the current Athlon XP at the same clock speed, on 32-bit code:

<A HREF="http://www.aceshardware.com/read_news.jsp?id=55000468" target="_new"> Check me out! </A>

This is probably for non-SSE2 code, SSE2 apps would be faster. While this is by no means going backwards in performance like the P4 did at launch, it's also not revolutionary, simply evolutionary.

The 486 gained about 60% IPC over the 386. The Pentium gained about 50% IPC over the 486. Those were revolutionary upgrades. The Pentium Pro actually went *backwards* in IPC on 16-bit code. After that it got difficult to compare clock to clock because the P2 was released at a faster clock than the Pentium or PPro, and similarly with the P3 and P4. The P4's just a totally different animal: in some apps (bandwidth hungry and SSE2) it gains 100%+ IPC over a P3, in others it loses 30% IPC. Nowadays it rules in the former apps by architecture and conquers the latter by brute speed.

Ritesh
 
It's true though, it's kinda like Episode 1. No matter how good Lucas made it, there's no way he could keep up with the hype. Everyone is now let down, so Episode 2 is a lot easier. Nothing to do with the quality of the product, everything to do with people's expectations.

<font color=blue>Hi mom!</font color=blue>
 
You're forgetting that AMD has publicly stated that Clawhammer is expected to be approximately 20-25% faster than the current Athlon XP at the same clock speed, on 32-bit code:
Click to expand...

So did Intel, I don't have a link but I do remember when they demoed P4 @ IDF or where ever they said how much higher preformance it will give etc.

nForce was the Same and we are seeing similar stuff from AMD. It's all Hype for now. We will have to see when it's released. Until then believe who ever you want to but it will still by Hype or Wishful thinking.

No company is ever going to say that their upcoming new product is not as good as their previous product. That's not the way you do business.

KG

"640K ought to be enough for anybody." - Bill Gates.
 
It's not about the overall performance I am going at, it's to prove there are such skim chances like an HSF falling off, that AMD would strip the Hammer like P4.
My point still stands, and if it changes, I'll be damned forever. Unless someone wants to argue why it would perform worse?

--
Luke, I am your father...but due to a bacon-slicing accident, your mother... :lol:
 
But ATM, texas_techie is the nearest AMD contact we can get, and he seems to hint all is going well, and that most likely it should hit PR3000 and more by its release. Remember, it's PR3000 or nothing. It's not like Intel with MHZ. 1.4GHZ may be powerful or crappy, while PR is "simulated" but much more accurate and cannot modify itself unless AMD wishes, and even then, to what chip would they go for? Athlon A?!

--
Luke, I am your father...but due to a bacon-slicing accident, your mother... :lol:
 
K8 is already demonstrated, it is smaller than a 0.18m AthlonXP, possibly smaller than Northwoods too, I forget.
Click to expand...

I agree K8 will be smaller then K7 but it will be larger on the same process. it will be smaller then XP since XP is using .18u and K8 will be using .13u. Also It will be smaller then NW, It's that what Jerry has been bragging about. The die size of the Hammer, how small it will be compare to Intel's processor.

KG

"640K ought to be enough for anybody." - Bill Gates.
 
It's not about the overall performance I am going at, it's to prove there are such skim chances like an HSF falling off, that AMD would strip the Hammer like P4.
Click to expand...

OK, I agree that there are still very slim chances that HSF will fall off. But I don't know what you mean by saying that AMD wouldn't strip the Hammer like P4. We don't know that for sure. I know hammer has been running and all but to same some money and to get it out on time the companies would do anything. For example, P4 had a less cache when it released. What if AMD did the same thing?

As far as I know all of the Hammer preformance that we here is wishful thinking, I think texas posted some benches which didn't show good numbers for hammer. But it is still in production.

KG

"640K ought to be enough for anybody." - Bill Gates.
 
Of course when it is in production, anything can pop up as bad. Remember the AGP not functioning back at IDF's intervention? Now it is fixed. AMD may sound like they are on high bumpy roads, but I am sure they will come out of it in the end, that's how it was before K7 came out anyway.

The thing is, if the A0 stepping was said to be one of the only times it ever ran THAT well for hours, stable and functional, even at low clock speed, it was a good stepping by AMD. I think they are happy of that, and will go on. Besides, since when would AMD opt for a less IPC, more MHZ version? I am not saying you said that, but that would be the end conclusion should AMD strip its CPU.

--
Luke, I am your father...but due to a bacon-slicing accident, your mother... :lol:
 
1.4GHZ may be powerful or crappy, while PR is "simulated" but much more accurate and cannot modify itself unless AMD wishes, and even then, to what chip would they go for? Athlon A?!
Click to expand...

You forget that PR ratings aren't that accurate. And yes, AMD is free to change them. To what? Hammer running 64-bit compared to Athlon running in 32-bit? If not, then everyone will scream that it's unfair. If so, then it's practically blatantly lying because nobody will be using 64-bit right off the bat. It'd be like Intel using Hyperthreading for benchmarks when the P4 was released.

<font color=blue>Hi mom!</font color=blue>
 
Yeah those benches I got sucked a$$.. but i know they have come a LONG way since then. The key here to remember that hammer isnt reinventing the wheel. If fact, IMO, the neatest thing about hammer isnt the CPU... its the mobo. Im looking forward to the extra bandwidth and lower latency. Of course, having a faster chip is always nice too :)
Anyway, i may have more stuff for you guys soon. Let ya know.

Benchmarks are like sex, everybody loves doing it, everybody thinks they are good at it.
 
I completely agree, the Hammer is way overhyped. Does that mean it'll suck? Not at all, but look how excited everyone was about the nForce, and how it really is. Similar thing.
Click to expand...


If via hadnt released the kt266a and stole the nforces thunder it would have met all of its hype, I mean 10-15% faster than any previous ddr solution????

The only reason it did poorly was like a week or 3 before via fixxed the kt266a and the nforce was too little too late.

As it stood on its own, the nforce is a blazing ddr chipset.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
 
Ritesh, 30% faster with more pipeline stages is very drastic, not only is it 30% faster per clock, but it can scale much further.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
 
As far as I know all of the Hammer preformance that we here is wishful thinking, I think texas posted some benches which didn't show good numbers for hammer. But it is still in production.
Click to expand...

Kemche texas's benches did show good for the hammer.


IIRC the claw got 800 points more than the same clockspeed axp with NOT A SUPER VIDEOCARD.

800 points at the same clockspeed can be a big deal.

WQhat makes you say they didnt do good?

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
 
Yes I know, but whatever the PR would stand for, if it is at PR3400, it has to be around it at least IF it doesn't go that accurate. Get my point? I am not saying the PR3400 will kick the ass, because that is hype, but I am saying that at such PR, I think AMD knows we are expecting nearby performance of such simulated MHZ, and that is why they simply cannot hide it by stripping in the end. OTOH Intel can say they will begin their P4 at 1.3GHZ, people will hype, but that speed can have different performance anytime, IPC is a big factor on that.

--
Luke, I am your father...but due to a bacon-slicing accident, your mother... :lol:
 
Besides, since when would AMD opt for a less IPC, more MHZ version?
Click to expand...
Ultimately, what does it matter which direction AMD goes? if they can provide a highly scalable CPU that has performance around PR3400 at the low end, does it realy matter if they had to crank up the clockspeed and sacrifice cpu efficiency to do it?

Now, I'm not saying that I want hammer to be gutted, or that it will be, but if performance is all that truely matters, then who cares how AMD gets there as long as they do?

Fleedip 2002, Microsoft's answer to bugs.
 
If via hadnt released the kt266a and stole the nforces thunder it would have met all of its hype,
Click to expand...

And if nVidia had released the nForce when it was originally supposed to, it would've been out before the KT266a by a good 5-6 months.

30% faster with more pipeline stages is very drastic, not only is it 30% faster per clock, but it can scale much further.
Click to expand...

I doubt two more stages will help it that much. The P4 is a 20 stage pipeline, the Hammer is only 12. But we'll see. SOI will help, of course.

<font color=blue>Hi mom!</font color=blue>
 
IIRC the claw got 800 points more than the same clockspeed axp with NOT A SUPER VIDEOCARD.
Click to expand...

That's just one test. And I hope the PR3400+ doesn't just use this test. It will use multiple benches. So one good number doesn't prove anything. Even P4 1.3 can beat any of the AMD's offering in Memory bandwidth and some SSE2 optimized apps.

Here's what texas said in this thread...

Yeah those benches I got sucked a$$.. but i know they have come a LONG way since then. The key here to remember that hammer isnt reinventing the wheel.
Click to expand...

May be the benchmarks don't look as bad now. But we still don't know that other then texas. We would have either Anand or Tom review the chip when they get some sample or when it releases. Until then the PR3400+ is just a hype and most likely a wishful thinking (again).

KG

"640K ought to be enough for anybody." - Bill Gates.
 
I agree with you man, i just read a newspaper acticle few days ago that the reporter had interview the director of star wars I and II. He said that because the I was so hyped before it comes out, everyone said it was very disapointing when it finally comes. so this time he lock almost all the info about II. It prove one thing "The higher the expectation the higher the disapointment." I am no AMDmeltdown but I personally think that the 30% increase is the highest increase in some programs. Not all programs will increase that much, i think if most of them increase 10 to 15% i will be happy. Its alll about the hype, history repeat itself too well; p4, nforce, kt266, the first p 60/66, and a lot more in sofwares world. So i think we shouldnt set the standard too high until hammer acturally comes out. Right now all we can do is guess with all the info we can find.
 
"The higher the expectation the higher the disapointment."
Click to expand...

Exactly what I'm saying. People are hyping themselves up for Hammer like it will revolutionize the way we do computers, end world hunger and enlarge your penis, but it's simply not going to.

And with Star Wars, I also think that some of the reason people were disappointed with Episode 1 is that they were comparing it with 4-6, not <i>just</i> 4. Can't compare one part of a trilogy to an entire trilogy.

<font color=blue>Hi mom!</font color=blue>
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom