Sign in with
Sign up | Sign in
Your question

Digitimes slams AMD's Puma announcement

Last response: in CPUs
Share
May 21, 2007 5:47:56 PM

Quote:

...it is questionable whether, when (or dare we even say if) it is launched, Puma will meet the hype AMD is currently trying to generate though these early announcements.

There is no doubt AMD is currently doing a fine job of creating PowerPoint presentations and press releases; but without any hardware to back things up, it is possible that the company will end up giving Intel yet another year to lead in the notebook market.




http://www.digitimes.com/systems/a20070518VL201.html

More about : digitimes slams amd puma announcement

May 21, 2007 6:19:59 PM

I came to the same conclusion when I read about Griffin and found that it is pretty much a tweaked K8 that will be the hope for AMD in 2008. As far as I can tell, between Griffin and the "Geode" processor, AMD is not doing that well in notebook land and seems to be falling further behind.
May 21, 2007 6:46:48 PM

Whats getting just as annoying, is people and the media, crying and complaining about AMD being hush hush about everything; then when AMD does speak up about forthcoming products and technologies, they get assaulted for not having an actual product.

Now in the case of Barcelona where they have workable samples and just talk about it, is justifiable. But things like Puma/Griffin need to relax, they're giving you the information your crying about, wait till they announce a sample date then moan and cry if they don't have samples ready then.
Related resources
May 21, 2007 6:49:34 PM

Hope the AMD Execs took a bonus out of the fire, errrrrrr...... bond sale two weeks ago because they wont be seeing one from sales for a long time.
May 21, 2007 8:23:51 PM

This is part of what we have been asking for from AMD what's in the pipeline for future production. The bigger question though is where are the benchmarks for the stuff that should be in production.
May 21, 2007 8:38:37 PM

Technologically it can be as advanced as the Centrino platform in total, but it will still lag. And I'm not talking about just performance.


By the way, why wasn't I informed of Ninja Celebration week? Since when was I tribute worthy?
May 21, 2007 8:39:29 PM

Quote:
I have a really good idea. Unless it is news of an actual release date of the currently delayed stuff or even better independent benchmarks of these parts, (K10, R600 midrange and xtx) can we please just stop with all the AMD news? All this kind of news is like beating a dead horse. We know AMD can not deliver on time and lately is all talk and no show. No need to remind us every day. No offense to the OP. Just these types of threads are starting to get old.


Words of wisdom...
May 21, 2007 8:54:30 PM

Quote:
By the way, why wasn't I informed of Ninja Celebration week? Since when was I tribute worthy?


I put a link in the LV Tribute thread in the "Other". Re-reading, it looks like your tribute is only half a week. :wink:

On topic - I agree with Jon Stokes (ARS) that using the K8 uarch is a good idea for now. No sense in having a quadcore-ready platform for notebooks ATM.
May 21, 2007 9:08:45 PM

Quote:
By the way, why wasn't I informed of Ninja Celebration week? Since when was I tribute worthy?


I put a link in the LV Tribute thread in the "Other". Re-reading, it looks like your tribute is only half a week. :wink:

On topic - I agree with Jon Stokes (ARS) that using the K8 uarch is a good idea for now. No sense in having a quadcore-ready platform for notebooks ATM.

K10 is about more than native quad core. There will be dual core variants. Using K8 as a base architecture suggests to me that 128bit SSE improvements, improved stack management and improved ooo processing are not going to be present. When you are trying to stay low power and run a slow clock rate, improvements to ipc are important. Maybe I'm not understanding what AMD is planning, but this is just plain bad as far as I can tell.
May 21, 2007 9:10:05 PM

Quote:
On topic - I agree with Jon Stokes (ARS) that using the K8 uarch is a good idea for now. No sense in having a quadcore-ready platform for notebooks ATM.


IMO, using K8 in mid-2008 is a disaster from a performance standpoint. By that time Intel will have fast Penryn-derived dual cores (significantly higher IPC than Merom) in the mobile space.

Why is AMD not choosing to use the dual core K10? This is the only way AMD has a chance at staying competitive in mobile in mid-2008. Something's going on here. Either AMD doesn't see enough K10 capacity by that time to supply mobile, or K10 is a power hog.
May 21, 2007 9:10:56 PM

Quote:
By the way, why wasn't I informed of Ninja Celebration week? Since when was I tribute worthy?


I put a link in the LV Tribute thread in the "Other". Re-reading, it looks like your tribute is only half a week. :wink:

On topic - I agree with Jon Stokes (ARS) that using the K8 uarch is a good idea for now. No sense in having a quadcore-ready platform for notebooks ATM.

Exactly. K8 is not slouch and if they can tweak these to run at say 2.6GHz, they will definitely be in the running. I would think though that Rana might be a good replacement as it is 10h but doesn't have L3.

I'm confident though because HP had the new 65nm Turions for sale right now and they were just announced as shipping this month.

Maybe that will convince the previous poster that AMD does keep their schedules pretty well. Things happen in the complex process of CPU manuf.
A company just released a DTR notebok with a 50W chip, so it maybe possible to use the higher clocked chips for gaming (Turion was the first SLI notebook - AlienWare).


Also, at one of these sites they actually had a pic of a running board so it's not vapor. I think it was Anand.
May 21, 2007 9:53:34 PM

Puma isn't designed to be a end-all, high performace platform. K8 is still a respectable choice for current applications - although I wouldn't recommend the Puma/Griffin combo for a desktop replacement, it should be good enough for a standard mobile platform.

Quote:
It made sense for AMD to stick with K8 for laptops, given the following factors:

Dual-core is ideal for laptops right now, so it makes sense to tweak an existing dual-core design instead of trying to adapt a new quad-core design.
Barcelona has some tradeoffs that are intended for 64-bit computing. You'll find plenty of 64-bit workloads in the server space, but not so many in laptop applications
Barcelona's microarchitecture also contains a few tradeoffs in favor of power versus performance. Again, this is more ideal for servers than laptops.
Griffin takes the basic K8 microarchitecture and "rewires" it for lower power. One aspect of this rewiring that AMD has revealed is that each of the processor's two cores has its own voltage plane. The separate planes enable the two cores to have their frequencies and voltages dynamically adjusted independently of one another so that power consumption can be tuned to fit the running workload on a per-core basis.

The I/O components on the die with the K8 core will also get tweaks and updates. Both the memory and HyperTransport controllers will be power-optimized so that they operate on a separate power plane from the processor cores. Again, this lets Griffin down-clock one core without slowing down I/O to the other core.
May 21, 2007 10:01:46 PM

Quote:
Maybe I'm not understanding what AMD is planning, but this is just plain bad as far as I can tell.


I think AMD is trying to get the most out of what they have. While I have no doubt that future notebooks based on K10 will perform much better, I don't think AMD has enough resources to push out a new uarch on the server, desktop and mobile fronts at the same time. What they can't afford to do is to short the supply channel again.

I think Puma is good for the everyday/value mobile platform (which is becoming the leading in sales out of all), while they get K10 rolling on the Server & Desktop end first. Note: I am assuming that Puma will be priced appropriately, of course.
May 21, 2007 10:05:47 PM

Quote:
Whats getting just as annoying, is people and the media, crying and complaining about AMD being hush hush about everything; then when AMD does speak up about forthcoming products and technologies, they get assaulted for not having an actual product.

Now in the case of Barcelona where they have workable samples and just talk about it, is justifiable. But things like Puma/Griffin need to relax, they're giving you the information your crying about, wait till they announce a sample date then moan and cry if they don't have samples ready then.


Thats because AMD is making claims, not giving numbers. What people want is '3dmark dragon: xxxFPS', not 'simulations show up to 40% gain over opteron in some aplications' or,'step B0 exceeded expectations'. What expectations, what applications, what simulations? AMD is blowing smoke, not giving facts.
May 21, 2007 10:08:51 PM

My God, It's full of stars...
May 21, 2007 10:19:35 PM

Quote:
Puma isn't designed to be a end-all, high performace platform. K8 is still a respectable choice for current applications - although I wouldn't recommend the Puma/Griffin combo for a desktop replacement, it should be good enough for a standard mobile platform.


K8 Puma may be fast enough to run today's applications but it has no hope of competing with Intel's Penryn variants on mobile, which will offer far better performance per watt. For every speed grade of K8 Puma, Intel will have a solution that performs better, consumes less power, and costs the same or less (due to smaller die size). Who in their right mind would buy Puma? As far as I can tell, Puma is dead in the water a year before launch.

Please explain your logic!
May 21, 2007 10:42:41 PM

If digitimes is right, and it takes AMD forever to release this - the you are absolutely correct. I am giving AMD the benefit of the doubt by assuming that they can get a tweaked K8 variant out the door before Intel has a mobile Penryn.

As I stated earlier, Puma/Griffin looks good for a current mobile platform - as long is it isn't intended to be a desktop replacement laptop. If AMD does dawdle & delay, then it doesn't have a chance - unless Intel prices their new Penryn mobile platforms too high (judging from pricing over the past year, I doubt this will be the case).

Don't get me wrong - I'm not in any way saying that this will be comparable to Penryn. I am just saying that if AMD can get this on the market now, it could be a good competitor to the C2D/Santa Rosa - again pending on price.
May 21, 2007 11:05:01 PM

puma is base on the dying k8 core. i dont know why AMD take this stupid move. they should have done it witht the kuma core(k10 dual core). furthermore, with amd habit of delaying products, i expect puma will only release end of next year. by that time, puma will look obsolete, just like Pentium 4 now. i dont think any consumer is interested with this old k8 core anymore.
May 22, 2007 12:20:59 AM



:twisted: :wink:


*Waits for the confusion.*
May 22, 2007 12:58:28 AM

... I think I just went bonkers.
May 22, 2007 1:05:23 AM

Quote:
Maybe Puma in reality will basically be just a Sempron type chip for low budget low power laptops. Good enough for say a regular college student that just wants it for writing papers and maybe music/movies. I would think these would be in like the cheap $399 laptops you can get @ wal-mart or even Dell.


This is the conjecture, which doesn't bode well for AMD. Shooting to make a processor worthy of being at Wal-Mart isn't a move up. It is just sad.
May 22, 2007 1:07:38 AM

Quote:
... I think I just went bonkers.


.....just... :?: :?: :?:
May 22, 2007 1:18:01 AM

Well I see Ninja appreciation week is going well. When do we have cake?

Back on topic, performance doesn't exactly win laptops. The stability of the platform and also its battery life does.

@Turpit
Since its all in my head, if I don't think I'm crazy, therefore, I'm not crazy. I think.
May 22, 2007 1:20:53 AM

Quote:

This is the conjecture, which doesn't bode well for AMD. Shooting to make a processor worthy of being at Wal-Mart isn't a move up. It is just sad.


Performance wise yes but considering most of the laptops bought are just those types then from a marketshare standpoint it could be good for them.

I think that by shooting this low that they are in danger of not being competitive at all. They don't have the economies of scale that Intel has to do the ultra-low cost chip with decent performance. They will get squeezed from the bottom by Core and Pentium M shrinks and from the top by Penryn. This product looks to be less promising than Turion.
!