Is Penryn going to finish AMD off once and for all?

[Hajpoj]

Then C2D came along and it was 40%(rough estimate) faster than AMD....


So when AMD finally starts shipping it's next gen CPUs, how are they going to overcome such an overwhelming performance gap in the face of Penryn?


I believe in order for AMD to stay afloat they must surpass Penryn significantly, not just keep up...

 

[MarcVenice]

clock for clock c2d is 20-25% faster then a amd athlon 64 chip. Since Intel blew past AMD, why shouldn't AMD at least be able to close the gap with Intel, and Penryn is a 'mere' die shrink, no ? Reported gains aren't that huge afaik.

If Intel can produce 4ghz chips though, and if Penryn is going to be as fast or just slightly slower then AMD's new cpu's, then 2-2.4ghz chips won't hack it.

 

[zsdersw]

The performance crown will come down to who can scale their architecture faster and to what degree. At launch, Barcelona and Phenom appear that they won't have the clock speed to surpass the Intel chips available at the time. Penryn will up and ante a bit on clock speed, but to answer your question.. no. Penryn won't finish off AMD.

If anything from Intel finishes off AMD as we know it, it'll be Nehalem and its derivatives.

 

[rmed64]

If Phenom is faster clock for clock, then they will be fine, even if they cant compete with the high clock speeds at launch.

 

[Cookie Monster]

Originally posted by: rmed64
If Phenom is faster clock for clock, then they will be fine, even if they cant compete with the high clock speeds at launch.

The thing is.. just how much faster is the phenom faster than the penryn clock for clock? Penryn will be more than 20% (maybe 30ish?) faster clock for clock than the AMD 64 X2s.

Not to mention that unlike AMD, these intel chips can hit VERY high frequencies.

If phenom doesn't do something special against the competition, this could be the sign of bad things to come in the CPU industry.

 

[coldpower27]

Originally posted by: rmed64
If Phenom is faster clock for clock, then they will be fine, even if they cant compete with the high clock speeds at launch.

This isn't Athlon 64x2 vs Pentium D, K8 is currently operating on a deficit in IPC compared to Core 2, and when Penryn derivatives come around your looking at this gap increasing.

K10 might be able to close the gap some, but I seriously doubt that Barcelona is going to be up the ante on IPC so much, that they could surmount Penryn's enormous clockspeed in comparison. Even K7 to K8 was only 20% or so on it's own and this isn't a revolutionary design, just evolutionary tweaks.

 

[sparks]

From an end user perspective, I certainly hope not. I don't know if you remember when the "budget" cpu was $300+, but I do not want to relive those days.

 

[Tboy012]

Originally posted by: sparks
From an end user perspective, I certainly hope not. I don't know if you remember when the "budget" cpu was $300+, but I do not want to relive those days.

i agree with you sparks... the market has to stay competitve so we can get good prices

 

[TuxDave]

Originally posted by: MarcVenice
... and Penryn is a 'mere' die shrink, no ? Reported gains aren't that huge afaik

Incorrect.

1) Penryn is not a 'mere' die shrink and performance gains are noticable in the applications they tested in the below article:

Comparing a 3.2GHz Penryn (1.6GHz FSB) to a 3.0GHz Conroe (1.33GHz FSB), Intel has measured more than 20% increase in gaming performance (with no code changes). For video encoding applications, if SSE4 is utilized, the same Penryn vs. Conroe comparison can offer more than a 40% increase in performance.

http://www.anandtech.com/cpuch...howdoc.aspx?i=2955&p=1

 

[MarcVenice]

So penryn will be more then 40% faster clock for clock compared to AMD's K8 chips. If so, AMD has to make a BIG step forward to be competitive, especially since they are coming out with chips between that top out at 2.4ghz. And I hope they do.

Btw I just read AMD will be releasing phenom in Q4, but only in limited amounts and different speeds. So it will be 2008 when it really hits the market.

 

[Mana]

1) Penryn is not a 'mere' die shrink and performance gains are noticable in the applications they tested in the below article:

Comparing a 3.2GHz Penryn (1.6GHz FSB) to a 3.0GHz Conroe (1.33GHz FSB), Intel has measured more than 20% increase in gaming performance (with no code changes). For video encoding applications, if SSE4 is utilized, the same Penryn vs. Conroe comparison can offer more than a 40% increase in performance.

The reason for the jump in performance there is the usage of SSE4 instructions. I would expect Phenom to have something similar.

Honestly though, I'm just going to wait for the real world benchmarks.

 

[JAG87]

AMD will not have 4ghz K10 chips, nothing even close to that. I will be very suprised if K10 even hits 3ghz. Although its 65nm, we are looking at the same overclocking capabilities of K8, which tops at around 3ghz with air. yet many reports say that K10 tops at around 2.6-2.8, but thats early samples so there is still time for another revision. thats definetely not enough to compete with 4ghz chips. and penryn will hit 4ghz with air FOR SURE. G0 hits 4ghz and its only 65nm with no HI-K. penryn is 45nm with HI-K.

just think about it, yorkfield extreme will release at 3.33 ghz, which is the higher than the 3 ghz of the QX6850. so if intel thinks they can up the frequency a notch, i think overclockers will find an even higher notch.

:laugh:

 

[coldpower27]

Originally posted by: Mana

1) Penryn is not a 'mere' die shrink and performance gains are noticable in the applications they tested in the below article:

Comparing a 3.2GHz Penryn (1.6GHz FSB) to a 3.0GHz Conroe (1.33GHz FSB), Intel has measured more than 20% increase in gaming performance (with no code changes). For video encoding applications, if SSE4 is utilized, the same Penryn vs. Conroe comparison can offer more than a 40% increase in performance.

The reason for the jump in performance there is the usage of SSE4 instructions. I would expect Phenom to have something similar.

Honestly though, I'm just going to wait for the real world benchmarks.

20% in gaming increase isn't from SSE4, Phenom is not likely to have anything higher then SSSE3, as AMD is usually 1 generation behind in implementing Intel instruction sets.

We probably will see SSE4 in Shanghai and it's derivatives.

 

[Hajpoj]

Originally posted by: coldpower27


We probably will see SSE4 in Shanghai and it's derivatives.

IF we see Shanghai.....

**runs and hides in secret underground bunker**

 

[BigDH01]

Originally posted by: rmed64
If Phenom is faster clock for clock, then they will be fine, even if they cant compete with the high clock speeds at launch.

Not quite true. That would place AMD in the budget segment like they are now. We know now that Intel is willing to keep the price war going as needed. AMD doesn't need to just match Intel's performance, they need to raise their margin or they're toast. If AMD can only compete in the budget to mainstream segment then expect Intel to practically give away Penryns at those price points. That would be very bad for AMD even if Phenom is comparable.

 

[undeclared]

HELLO PEOPLE
Amd and Intel don't survive from enthusiasts!!!!!!

They survive from Brand name desktop and laptops

Nothing will kill_EITHER_ of them.

 

[BigDH01]

Originally posted by: xtwells
HELLO PEOPLE
Amd and Intel don't survive from enthusiasts!!!!!!

They survive from Brand name desktop and laptops

Nothing will kill_EITHER_ of them.

Not true. Just having brand name customers such as Dell, etc, doesn't not guarantee the survival of your business. Just look at Q2 financial results from AMD. Their computing solutions revenue was up almost 200 million. Presumably, much of this income is derived from those large brands such as HP and Dell. However, their computing solutions segment still lost 258 million dollars, about a ~70 million improvement from the previous quarter. I'm not saying that AMD is going out of business, simply pointing out that sales to large manufacturers doesn't necessarily mean a successful business.

 

[Markfw]

Also, in the server market, they are not as far behind, and getting decent revenue I think.

 

[TuxDave]

Originally posted by: Mana

1) Penryn is not a 'mere' die shrink and performance gains are noticable in the applications they tested in the below article:

Comparing a 3.2GHz Penryn (1.6GHz FSB) to a 3.0GHz Conroe (1.33GHz FSB), Intel has measured more than 20% increase in gaming performance (with no code changes). For video encoding applications, if SSE4 is utilized, the same Penryn vs. Conroe comparison can offer more than a 40% increase in performance.

The reason for the jump in performance there is the usage of SSE4 instructions. I would expect Phenom to have something similar.

Honestly though, I'm just going to wait for the real world benchmarks.

What about the bolded statement in terms of gaming performance?

 

[bryanW1995]

I'm just stoked to see them talking about a 1.6 ghz fsb penryn chip. That puppy could hit 4.5 on air...sigh...I think I'm in love...

 

[VirtualLarry]

Originally posted by: bryanW1995
I'm just stoked to see them talking about a 1.6 ghz fsb penryn chip. That puppy could hit 4.5 on air...sigh...I think I'm in love...

Yeah, but look at the FSB you would need to OC it. I'm not liking 1.6Ghz FSB chips for that reason.

 

[bryanW1995]

It's not so bad if you have ddr2 1066...just at stock that will get you a 33 % bump. if you get a small ram oc to 1126 you can get 563x8 = 4504. Most high end p35 mobos will run from 500-550 right now with shipping bios. These boards are designed for penryn so its reasonable that they'll go even higher in the next 6-12 mos. So if you have, for example, an abit ip35 pro with crucial ballistix pc 8500, you are more likely to be capped by the processor in any overclocking situation that you're likely to see.

 

[lenjack]

Even if you're an Intel fan, you need AMD to stay around. As mentioned before, An Intel without AMD competition will result in much higher cpu prices and much slower cpu advances. We need both of them to do well.

 

[Cookie Monster]

Originally posted by: lenjack
Even if you're an Intel fan, you need AMD to stay around. As mentioned before, An Intel without AMD competition will result in much higher cpu prices and much slower cpu advances. We need both of them to do well.

Quite true. Look what AMD did to intel when they walloped intel's netburst architecture. It created the devils advocate, the intel conroes.

Now is AMDs turn to step up a notch or two.

 

[bryanW1995]

I really want to see amd be successful. As I've mentioned before, I've bought 6 straight amd cpus. I still have 4 of them in fact. I just don't see a way for amd to compete in the high end right now. Barcelona/phenom MIGHT be as much as 10% faster than penryn clock/clock. Even the biggest amd fanboy is unlikely to predict anything higher than that. Penryn is being introduced at 3.33 and will almost certainly get very near 4 ghz if not all the way there. Is phenom going to get to 3.6 before nehalem comes out? I think that 3.0 is very optimistic in the next 15 months, then nehalem will bust out the whooping stick on them again. Can fusion save amd in 2009? Something had better change quick or we will go back to the budget $300 cpu.