AMD's answer to the i7?

legcramp

[H]F Junkie
2FA
Joined
Aug 16, 2004
Messages
12,439
OK, we are revealed Intel's new cpu's the Core i7 chips which are very fast in general.

What does AMD have that's coming to counter this? Maybe something like the Athlon 64 debut owning the Pentium 4's? :(
 
AMD has the Deneb and Shanghai processors coming out but I wouldn't expect them to beat the I7s.
 
I don't see AMD competing in the high end for a long time to come. Whats worse, whatever bottlenecks Intel had with C2D in multi-socket servers will be a thing of the past with i7, so the performance advantage AMD currently has in that market will soon be at an end as well.
 
Maybe the uarch post-Bulldozer will catch up to i7. ;)
 
Fusion is aimed at the lower end market. :p And it's just a K10-based CPU + graphic chip in the same package (multi-chip module) anyways.
 
AMD has the Deneb and Shanghai processors coming out but I wouldn't expect them to beat the I7s.

I would be surprise if AMDs Deneb and Shanghai processors could be the Core 2 series. But I hope they do! I want an AMD come back but then reality sets in =(
 
AMD's answer will be price war.
Intel will in turn match the price war in the low to mid sectors while probably charging a little bit more because they will expect the chips to sell a little better based on name and people hearing other saying they are the fastest (Referring to the top end) and generalizing to their entire product line. Meanwhile on the top end Intel will charge a fortune for the chips out of AMD's competitive range ;) there you have the next few years of computing!
 
AMD didn't have any answer for nVidia's G80 either :rolleyes:

At least someone is keeping the faith! :p

The difference is that CPUs have a far longer development cycle than GPUs. Assuming Deneb doesn't measure up to i7 (unless you know something we don't) it'll take an architectural overhaul from AMD to catch i7, aka Bulldozer. The only problem is that by the time Bulldozer is supposed to launch Intel will be on to their next 'tock' already.
 
AMD's answer will be price war.
Intel will in turn match the price war in the low to mid sectors while probably charging a little bit more because they will expect the chips to sell a little better based on name and people hearing other saying they are the fastest (Referring to the top end) and generalizing to their entire product line. Meanwhile on the top end Intel will charge a fortune for the chips out of AMD's competitive range ;) there you have the next few years of computing!

Dude, that sounds exactly like the past 2 years post Conroe... :eek:
 
i wish. im a open minded person and i couldnt care less wwho made the latest and greatest product weather its AMD, intel or microsoft i do not care but i care about competitions as if AMD or another company had a cpu that rivals an i7 then these x58 mobo's wouldnt be soo expensive
 
At least someone is keeping the faith! :p

The difference is that CPUs have a far longer development cycle than GPUs. Assuming Deneb doesn't measure up to i7 (unless you know something we don't) it'll take an architectural overhaul from AMD to catch i7, aka Bulldozer. The only problem is that by the time Bulldozer is supposed to launch Intel will be on to their next 'tock' already.

i7 > "deneb"...search the web, only a crackpot would assume a change from 65nm to 45nm can suddenly make a CPU prefomce by a factor of x2
 
i7 > "deneb"...search the web, only a crackpot would assume a change from 65nm to 45nm can suddenly make a CPU prefomce by a factor of x2

To be fair, it's not just a die shrink. There should be a few tweaks to the architecture, but yeah, it won't change its fate when compared to i7 and there's still the question if it's going to be able to catch up with Penrym.
 
and there's still the question if it's going to be able to catch up with Penrym.
From early Deneb benchmarks, the answer is no compared to Yorkfield. It may manage to finally catch up to Kentsfield in some areas Agena lagged.
 
Almost two years after G80's launch, but yeah...they did.

By then they were competing with the Geforce GTX 280. The 4870 is a little slower but wins on price. The 4870 X2 is faster, but is more expensive.

AMD has the Deneb and Shanghai processors coming out but I wouldn't expect them to beat the I7s.

I wouldn't expect them to do all that well compared to Yorkfield.

I don't see AMD competing in the high end for a long time to come. Whats worse, whatever bottlenecks Intel had with C2D in multi-socket servers will be a thing of the past with i7, so the performance advantage AMD currently has in that market will soon be at an end as well.

Absolutely. I agree with you completely. I think the Core i7 will really come into its' own on the server side of things. Its' a great desktop processor no doubt but it will really shine in the server market.

Maybe the uarch post-Bulldozer will catch up to i7. ;)

Right. So far I don't think any of us has heard of anything that will give AMD the ability to equal Yorkfield, or Nehalem anytime soon. Bulldozer is their best hope to level the playing field.

AMD didn't have any answer for nVidia's G80 either :rolleyes:

No they didn't. It took them two years to come out with something worth while and by then they were competing with the Geforce GTX 260 and Geforce GTX 280.

I would be surprise if AMDs Deneb and Shanghai processors could be the Core 2 series. But I hope they do! I want an AMD come back but then reality sets in =(

AMD is going to need more than a die shrink and a couple of tweaks to match up to the Core 2 series. Competing against Nehalem is another matter entirely. I just don't see it happening.

i7 > "deneb"...search the web, only a crackpot would assume a change from 65nm to 45nm can suddenly make a CPU prefomce by a factor of x2

Quite right. It just isn't going to happen. I know many AMD Fans and loyalists have high hopes for Deneb but frankly, its' totally unrealistic to think that a die shrink and a few tweaks are going to suddenly make what was essentially kind of a lame duck in the first place run well enough to compete against processors that Agena failed to be competitive with. Deneb just isn't likely to be different enough to get the job done.

From early Deneb benchmarks, the answer is no compared to Yorkfield. It may manage to finally catch up to Kentsfield in some areas Agena lagged.

And this is what I think amounts to the best AMD can hope for at this point. It boggles my mind how Phenom could even have been launched in the first place. Its' slower than the Athlon X2 much of the time or only marginally faster. It doesn't really overclock all that well and the stock clock speeds aren't impressive to say the least.

The thing is that now with Core i7 the only thing AMD has going for them is the price. I remember how during the Athlon 64 days several people (including myself) said Intel systems felt smoother and more responsive to use. We generally attributed that to Hyperthreading though the motherboard chipsets may have had something to do with it in all reality. For whatever reason Intel's platform was superior for general usage. Now many AMD fans today were saying that AMD Athlon X2 and Phenom systems run smoother than their Intel counterparts and are more responsive despite losing at just about every benchmark out there. Now with the Core i7 you have Hyperthreading and QPI which is the same thing essentially that Hypertransport is. Wich takes all the possible advantages away from AMD. Now AMD fans and loyalists can only cling to price because there are no advantages on a technical level. I don't see how any technical advantages could even be perceived (like the unsubstantiated claim that Phenom systems feel smoother and more responsive than Core 2 Quad systems) even by the most hardcore fans.

It is indeed a dark time for AMD. At least on the processor side.
 
By then they were competing with the Geforce GTX 280. The 4870 is a little slower but wins on price. The 4870 X2 is faster, but is more expensive.



I wouldn't expect them to do all that well compared to Yorkfield.



Absolutely. I agree with you completely. I think the Core i7 will really come into its' own on the server side of things. Its' a great desktop processor no doubt but it will really shine in the server market.



Right. So far I don't think any of us has heard of anything that will give AMD the ability to equal Yorkfield, or Nehalem anytime soon. Bulldozer is their best hope to level the playing field.



No they didn't. It took them two years to come out with something worth while and by then they were competing with the Geforce GTX 260 and Geforce GTX 280.



AMD is going to need more than a die shrink and a couple of tweaks to match up to the Core 2 series. Competing against Nehalem is another matter entirely. I just don't see it happening.



Quite right. It just isn't going to happen. I know many AMD Fans and loyalists have high hopes for Deneb but frankly, its' totally unrealistic to think that a die shrink and a few tweaks are going to suddenly make what was essentially kind of a lame duck in the first place run well enough to compete against processors that Agena failed to be competitive with. Deneb just isn't likely to be different enough to get the job done.



And this is what I think amounts to the best AMD can hope for at this point. It boggles my mind how Phenom could even have been launched in the first place. Its' slower than the Athlon X2 much of the time or only marginally faster. It doesn't really overclock all that well and the stock clock speeds aren't impressive to say the least.

The thing is that now with Core i7 the only thing AMD has going for them is the price. I remember how during the Athlon 64 days several people (including myself) said Intel systems felt smoother and more responsive to use. We generally attributed that to Hyperthreading though the motherboard chipsets may have had something to do with it in all reality. For whatever reason Intel's platform was superior for general usage. Now many AMD fans today were saying that AMD Athlon X2 and Phenom systems run smoother than their Intel counterparts and are more responsive despite losing at just about every benchmark out there. Now with the Core i7 you have Hyperthreading and QPI which is the same thing essentially that Hypertransport is. Wich takes all the possible advantages away from AMD. Now AMD fans and loyalists can only cling to price because there are no advantages on a technical level. I don't see how any technical advantages could even be perceived (like the unsubstantiated claim that Phenom systems feel smoother and more responsive than Core 2 Quad systems) even by the most hardcore fans.

It is indeed a dark time for AMD. At least on the processor side.

Long post but i think amd is not doomed; will it have an answer to the i7 yes thats bulldozer/fusion.

will amd's 45nm chips compete with the kentsfeild it will with out a problem amd still has the advantage in servers till the dp i7's are out

will it be completive with the i7 it is a possibility if it can scale and from everything i have seen it can and if the price is right [300x2 for a chip and board is insanity] (0 sec tech news junkie here)

amd did not just shrink and tweak the failed k10 they redesigned it and made it better hell its coming out this month rather than in q1 of 09 thats a good sign

in two weeks we will see so do not count amd out i find that an editor at [H]ardocp trashing an unreleased product and being negative towards it with little research kinda funny
 
amd did not just shrink and tweak the failed k10 they redesigned it and made it better hell its coming out this month rather than in q1 of 09 thats a good sign

Proof or I call FUD...
 
Long post but i think amd is not doomed; will it have an answer to the i7 yes thats bulldozer/fusion.

i never said AMD is doomed. If anything, AMD has proven time and time again that they can continue to operate without actually making any money.

will amd's 45nm chips compete with the kentsfeild it will with out a problem amd still has the advantage in servers till the dp i7's are out

Just how are AMD's 45nm chips going to compete with Kentsfield? A die-shrink doesn't improve performance. Yes I'd expect higher clock speed processors to be released as a result of the shrink but frankly, as it stands today Phenom needs about a 200MHz clock speed advantage just match a Core 2 Quad. 400MHz stock clock increases seem unlikely out of the 45nm chips. It isn't out of the question by any means, but still, I think you are reaching here. As far as the enthusiast is concerned the stock clocks aren't really an issue since they don't typically buy the highest clock speed models anyway. If Deneb can't overclock enough to beat the clock speed gains that you can get from a Core 2 processor then Deneb will be a loser in the eyes of the enthusiast.

will it be completive with the i7 it is a possibility if it can scale and from everything i have seen it can and if the price is right [300x2 for a chip and board is insanity] (0 sec tech news junkie here)

Considering AMD hasn't yet even matched Kentsfield or Yorkfield, there is NO possible way that Deneb is going to match or even compete with Core i7. That just isn't realistic. When you have lower IPC from the start, linear clock speed scaling when you are at a clock speed deficiency already won't get the job done.

amd did not just shrink and tweak the failed k10 they redesigned it and made it better hell its coming out this month rather than in q1 of 09 thats a good sign

Everything I've read on the subject points to Deneb being a simple die-shrink with an increased amount of cache. Just what improvements do you think AMD made with Deneb compared to Agena? What specific architectural differences have you saying this? What are your sources?

in two weeks we will see so do not count amd out i find that an editor at [H]ardocp trashing an unreleased product and being negative towards it with little research kinda funny

I'm a realist. Nothing more. I'm not trashing anything. I'm simply facing facts. It would take a massively different design than what they've got now for AMD to truly compete against Yorkfield and against Nehalem. Matching Kentsfield? Maybe. There is just no credible information out there at this point that lends any evidence to the possibility that Deneb would be improved enough to match Yorkfield or Nehalem. I'd love to be proven wrong. I'd love for AMD to be more competitive so that we can have a benefit from a good old fashioned price war again. I just don't see it happening.

As far as "research" goes the only "evidence" that Deneb is going to be massively superior comes from AMD fanboy's who are posting on forums that they've heard from someone who heard from a guy's roomate that knows an AMD engineer that says that Deneb will be x2 as good as Agena. Let's see some credible sources and I'll look at the information. If I'm wrong I'll admit to it. Still based on the evidence thus far from credible sources I don't think that I'm off base at all. If I am I don't think its' by much.
 
Deneb at 45nm might come close to a Core 2 performance level, but I am not holding my breath. As for it catching Nehalem. Well shit in one hand and hope in the other, and let's see which one of YOUR hands fills up first.
 
i'm now working for a company that doesn't have a real answer to the competitor's current and future offering :(

and watch out for more bad news.....I guess sooner or later it will leak out here in the internet.
 
Single threaded applications or multithreaded? Single threaded Core 2 and i7 is similar

Phenom is better on multithreaded applications now comparing to Core 2, Core 2 is only better on multithreaded applications that have been optimized for Core 2, Core 2 market share is big, applications that need speed need to do this. Now that i7 is out Core 2 will soon be geting worse results becaise developers don't need to think about Core 2

Core 2 is faster than Phenom clock for clock. It doesn't matter if the application is multi-threaded or not. Face facts, Phenom is not competitive with current Core 2 offerings and is certainly no where near the ballpark of Core i7.
 
Core 2 is faster than Phenom clock for clock. It doesn't matter if the application is multi-threaded or not. Face facts, Phenom is not competitive with current Core 2 offerings and is certainly no where near the ballpark of Core i7.
You should read more about processors and how they work ;) They are fast in different areas, Core 2 lives on its fast and big L2 cache. Depending on how the applications is done this is more or less good for performance. Browsers, databases then the Core 2 don't have a chance.
 
You'd know better than I.

As I understand it, unless they dramatically redesign the architecture (which isn't even remotely realistic in this stage of the product cycle) I just can't see how it would match Nehalem. I don't think matching Kentsfield is totally out of the question, and matching Yorkfield may or may not be a bit of a stretch.

Keep in mind that Phenom and Core 2 are pretty much neck-and-neck at 2.66GHz, it is when we see scaling above 3GHz that Core 2 runs off.

Single threaded applications or multithreaded? Single threaded Core 2 and i7 is similar

Phenom is better on multithreaded applications now comparing to Core 2, Core 2 is only better on multithreaded applications that have been optimized for Core 2, Core 2 market share is big, applications that need speed need to do this. Now that i7 is out Core 2 will soon be geting worse results becaise developers don't need to think about Core 2

Please share proof to back up your statements.
 
I'm going to guess here hope! heh since you can hold an infantile amount of hope in one hand and only so much shit in the other *waits for the nda to lift*


I am not under NDA with AMD nor has it supplied us with any samples of 45nm parts. I figured they would surely have done that for our Nehalem launch so we could have put it into the comparison. That is a HUGE opportunity to miss. That leads me to think a few things about Deneb. I was a huge proponent of Phenom and heard how good it was from AMD insiders for literally years. I ain't buying it till I see it this time around.
 
Most review sites are PRO Intel so it isn't strange that AMD keeps the processors until launch. Just check the types of applications that is used to test i7, most tests suddenly focus on multithreaded performance and/or memory intensive apps. Why hasn't this been done before testing phenom?
When did you see some review site testing database performance, encryption, browser or games using a slow videocard to filter out the lowest FPS values?

I am totally missing your point.

What sites are "PRO Intel" exactly?

AMD has been giving us early processors for years. That does not mean we are writing stories about them.

We focus on single threaded and multithreaded applications and have been doing so for years now.

As for memory intensive apps, please let me know which ones these are.

We have been testing multithreaded apps that we could identify since dual core hit the market actually even before with HyperThreading.

Don't know how to answer your last question. But you sort of sound like you got the tinfoil hat on today.
 
That's true. However it is rediculously easy to get Core 2's over 3.0GHz and not nearly as easy to do this with the Phenom. Plus there are already quite a few Core 2 models over 3.0GHz at stock speeds. Granted most are dual core, not quad core.

Er, OK. My point is that Deneb had better have some freakishly better scaling than Phenom up to an beyond 3GHz and I just don't see that happening unless AMD has made massive core changes.
 
Oh geez, he's back spreading fud again.

Reviews have ALWAYS been important. "the right" applications? Something tells me, any application that favors Intel (which would be atleast 90% of them) would be considered the "wrong" application as far as you're concerned. Intel > AMD period, end of story, nothing you say will change that. Get over it already.
 
Oh geez, he's back spreading fud again.

Reviews have ALWAYS been important. "the right" applications? Something tells me, any application that favors Intel (which would be atleast 90% of them) would be considered the "wrong" application as far as you're concerned. Intel > AMD period, end of story, nothing you say will change that. Get over it already.
How many are using CrossfireX or 3-sli? 0,01 % ?
 
Kyle/Dan = 1
kassler = 0

Sigh. kassler, you're not giving explicit examples of your argument. No one is bashing you (yet), but this site and it's community is not built on generalized claims of fanboy-isms, and we don't tolerate loyalties with no foundation for them. Please cite some references to your arguments...and what the hell does Crossfire and Crossfire X have to do with this argument, at all?

On a side note, as a long standing AMD partner, we have heard information on an upcoming 12 core (per socket) server CPU code named Magny-Cours (thank you for that...I couldn't recall the spelling), to come after Deneb. Initial projected launch I believe is 2H 09 or Q1 2010, but the dates keep changing. The rest of the details either escape me, or were not initially disclosed. We're on a list to get some engineering samples of the next few sets of server CPUs that AMD is producing, since we deal very heavily with the National Laboratories...and those boys only like to break stuff (well, LANL more than Sandia, anyways). Should be interesting, but I have very few details at this point. We are kind of just waiting to hear something, for our part of it.
 
AMD would still have a chance if their new CPU is cooler and can reach a higher clock. Looks like Core i7 is quite hot so this could limit the OC when compared to the C2Q.

Looks like using an integrated memory controller does increase the power consumption of the CPU after all. Some people didn't believe me when I said that the Phenom is hotter because of the imc and they said that AMD 65nm production is bad. Imo, the temperature on the 45nm Core i7 is not that impressive either.

Btw Kyle, I'm sorry if I miss it, but I didn't see any power consumption test on the Core i7 from [H] yet.
 
Sigh. kassler, you're not giving explicit examples of your argument. No one is bashing you (yet), but this site and it's community is not built on generalized claims of fanboy-isms, and we don't tolerate loyalties with no foundation for them. Please cite some references to your arguments...and what the hell does Crossfire and Crossfire X have to do with this argument, at all?

I was just explaining one reason why AMD could avoid sending processors to review sites. If you like Intel you will of course don't understand that reason.
Regarding this site so if you criticize (comparing AMD to Intel) Intel in the Intel area you will be warned, but in the AMD area it is possible for Intel fanboys to bash AMD as much as they want.
 
Back
Top