Columns A Phenomenon in Warsaw

[Tim S]

http://www.bit-tech.net/columns/2007...on_in_warsaw/1

Last week, I spent some time with AMD in Warsaw, Poland with the intention of finding out what's going on inside the newly-crowned platform company. The three elements that make up the Spider's web were the talk of the town, but I left somewhat disappointed.

[Cupboard]

Great article.

Its a shame that these chips cannot compete with the high end Intel parts, but it seems that at the mid to low end they have a lot going for them. Someone with little upgrade money could easily buy one part per year and still have everything working together because of the backward and forward compatibility of everything. For me, it would be and ideal solution, to be able to move with the times, without having to replace everything that is still pretty good which is rather a waste.

On a different note, I suppose that if they are having problems with these processors at high clock speeds, overclocking is going to be rubbish?

[Tim S]

Originally Posted by Cupboard

On a different note, I suppose that if they are having problems with these processors at high clock speeds, overclocking is going to be rubbish?

For the most part, everything I've seen so far points to poor overclocking on the initial chips. Once the process matures though, I would hope that speeds start to pick up pretty quickly - especially with Penryn already in-hand at 3.2GHz.

There are some select samples out there clocking well, but I believe that most of the volume out there won't clock all that well - I'm sure if there was a lot of headroom in the chips, AMD would have launched at 2.4GHz in volume.

[Woodstock]

as an amd fan, i was cringing all the way through that article.

[mrb_no1]

i was an AMD fan, and couldnt wait any longer to upgrade so went to intel with the core 2 duo and overclocked it....whilst i would have preferred to stay with AMD, i cant see me going back to them in the near future. intel keeps upping the bar and AMD is failing. What they need to win the hearts and confidence of some enthusiasts is launch a top line gpu or cpu that can compete/beat a top intel chip as for me looking at amd they look like a half hearted company that made some headway once, whilst the p4 was doing intel no favours but failed to capitalise on the situation and are now dithering about with slower clocked cores and weak marketing that will probably lend itself to being officially targeted at the lower end of the spectrum of users. Its a damn shame that i have to agree with Tim's 8 ball, the industry needs some fierce competition, i think thats what made intel pull its finger out its a$$ after the pentium4 fiasco to come up with the core arcitechture etc etc, needs to work the other way now though! pah

peace

fatman

p.s nice article Tim

[[USRF]Obiwan]

I remember a Vaporware console named "Phenom" or was it "Phantom" I cant remember.

Anyways...

i loved this part: "I was greeted by some rather puzzled looks from AMD’s execs". I bet this was the first time they actualy heard what "we" thought about the whole s939 to am2 upgrade disaster. There was no performance gain at all. Same proc new feet, same memory, same everyhthing really with a new socket. And then count the motherboard releases over the last 2 years, Intel 775socket mobos around 700+ different types/brands against 50+ am2 mobos. Did not see any new improved/better/feature packed am2 mobo for over a year... not even a am2+

[Shadow_101]

Good Read.

Although I wouldn’t say I’m a fanboy, I’ve always tried to use Green team. I’m disappointed that there not brining out anything special anymore. I cant see my next system being a AMD

[[USRF]Obiwan]

Originally Posted by Shadow_101

Good Read.

Although I wouldn’t say I’m a fanboy, I’ve always tried to use Green team. I’m disappointed that there not brining out anything special anymore. I cant see my next system being a AMD

I agree, my last intel system was a pentium 133. I always looked at the bang for the bucks. And right now its Intel all the way.

[samkiller42]

Very interesting read then Tim, thanks.
If you personally were in the situation where you need to upgrade, would you go Intel or AMD, based on the current reports that are about?

Sam

[Tim S]

It depends what you're after Sam, but I still think the Q6600 represents fantastic value for money, even though the Phenom 9700 is about the same speed as it. By the time the 9700 is available, we'll have the Q9450 (2.66GHz Penryn) at the same pricepoint as the Q6600.

[Nictron]

"That’s a good thing – as AMD isn’t suddenly going to alienate the AMD-faithful this time when they go out to upgrade their current AM2 systems in the next year and a half."

This might be there only salvation at this point, the fact that you can upgrade to a good performing quad core without any other upgrades needed!

[Brooxy]

I've not used an Intel in my main rig for about 5 years now, even my first PC was running an AMD chip (Am5x86).

Thinking when upgrade time comes, it might be the end of an era. AMD are trying, but just can't keep up, let alone overtake at the moment.

[Shadow_101]

Originally Posted by [USRF]Obiwan

I agree, my last intel system was a pentium 133. I always looked at the bang for the bucks. And right now its Intel all the way.

Indeed my last Intel was 166!

[cjmUK]

I'm afraid I lack loyalty... I've had Intel, Cyrix ad AMD CPUs over the years. Currently on C2D.

I'd like to see AMD back in the game though because competition keeps everyone on their toes. I don't want Intel to get complacent. We need Phenom and it's successors to be reasonably worthwhile, to make sure C2D/C2Q remain fast and remaon cheap.

[JumpingJack]

Tim, reading the article we all can sense your frustration and disappointment. As a matter of correction to ensure the facts are there, AMD does not support SSE4 insructions, only 4 of the 41 are available on this CPU, AMD calls it SSE4A but it is not the complete instruction set.

[Tim S]

Originally Posted by JumpingJack

Tim, reading the article we all can sense your frustration and disappointment. As a matter of correction to ensure the facts are there, AMD does not support SSE4 insructions, only 4 of the 41 are available on this CPU, AMD calls it SSE4A but it is not the complete instruction set.

Hey Jack, thanks for this - while I was at the event, AMD referred to it as "SSE4" support but having just gone through AMD's technical documentation for Phenom, you're right that it's referred to as SSE4a. I'll correct that part of the article now.

[kickarse]

Who would have guessed the Phenom would be less than Phenom... spectacular?!

[E.E.L. Ambiense]

Yeah. That was rather painful. Like many, I switched back to Intel because AMD just wasn't upping the bar. And the whole 939 thing did kinda peeve me off, but hey. That's their choice. I truly hope they do get something going soon though.

[JumpingJack]

Originally Posted by Tim S

Hey Jack, thanks for this - while I was at the event, AMD referred to it as "SSE4" support but having just gone through AMD's technical documentation for Phenom, you're right that it's referred to as SSE4a. I'll correct that part of the article now.

Tim, thanks for ensuring it is clarified ... AMD has recently been using loose wording and slight of language to push features that are oddly different than percieved, it is important the press ensures an accurate representation to avoid confusion... classic example is their ACP metric, I have seen in several cases people using the ACP interchangably for TDP.

Not quite as critical, you mention it is 33% wider... not sure if you are referring to the issue width and, frankly, I have not seen any official statemtent from AMD -- however, I was under the impression that Barcelona/Agena based on this core was still 3-issue, if that is true then it was not technically widened. They did increase the depth of the re-order buffer, which I find interesting.

Interesting, because it contributed significantly to an IPC gain... what this means is that a deeper reorder buffer is finding more instruction level parallelism in the x86 code base ..... important, because if Intel should choose to deepen the buffer then some gains can be expected ...how much I do not know, as other architectural enhancements also contribute to the IPC gain....

Again, thanks for adding the detail to a well written opinion piece.

Jack

[Shatylle]

I too as an avid AMD supporter am a bit disappointed as I was hoping for more benefit to appear from the different approach to architecture over the Intel line, and maybe there will be some still to be seen in later testing. But I am not going to be switching any time soon and I did make my choice to stick with the AMD processors after the rise of the Duo core, and bought a Crosshair board so I am not sorry to hear that they will be much compatibility to come socket wise. At the end of the day raw speed is not of such a concern as stability and in this regard I have never had any problems with my AMD machines. Bang for buck wise I am currently running my 4200+ at 2.7 stable so I am happy there.

It is my believable that the primary cause of AMD's problems is the fact that for over two years they have been reacting to Intel products and I suspect messing with there design and production timeline to compete, also in the mix is there purchase of ATI which I see them spending too much time trying to restructure and integrate into there future plans, this leaves an already underpower business wise AMD trying to spread itself too thin too fast.