Aceshardware

(not so) temporary home for the aceshardware community
 FAQ •  Search •  Register •  Login 
It is currently Thu Apr 24, 2014 12:21 am

All times are UTC + 1 hour



Welcome
Welcome to <strong>Aceshardware</strong>.

You are currently viewing our boards as a guest, which gives you limited access to view most discussions and access our other features. By joining our free community, you will have access to post topics, communicate privately with other members (PM), respond to polls, upload content, and access many other special features. Registration is fast, simple, and absolutely free, so please, <a href="/profile.php?mode=register">join our community today</a>!


Post new topic Reply to topic  [ 91 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6, 7  Next
Author Message
 Post subject:
PostPosted: Sun Jul 29, 2007 5:18 pm 
diep wrote:
My interest is whether one can keep 1 execution unit busy nearly nonstop executing multiplies 64x64 integers (giving a 128 bits result and it needs all 128 bits) and the other execution units executing other stuff, also SSE2 (64 bits integer).

I think that you can't choose what execution unit will be non-stop busy. That is determined by internal organisation of CPU. Execution units executing internal RISC microcode and reservation stations and reorder buffers fill out execution resources. CPU manufacturers are optimized CPU very well for optimal superscalar execution.


Top
  
 
 
 Post subject:
PostPosted: Tue Jul 31, 2007 5:08 am 
Offline

Joined: Sun Jul 22, 2007 11:06 am
Posts: 259
IdRatherNotSay wrote:
Some of you guys are funny.

I also fail to see the surprise; I told you before AMD had silicon running near 3GHz when Charlie misinterpreted that and thought it meant AMD would launch Opterons at that speed. Its no surprise AMD shows off one those cherry picked chips. Dont expect them to able to actually manufacture any quantity at this point though.


Actually, that is not why I said it. I based my comments on very specific information that is not public yet.

-Charlie


Top
 Profile  
 
 Post subject: I beg to differ
PostPosted: Tue Jul 31, 2007 5:09 am 
Offline

Joined: Sun Jul 22, 2007 11:06 am
Posts: 259
IdRatherNotSayEither wrote:
That's not a cherry picked part, by the way


I was told it was directly, but I was also told it was not heavily sorted. The take home message I got was 10s tested, not thousands.

-Charlie


Top
 Profile  
 
 Post subject: This is a flat out lie
PostPosted: Tue Jul 31, 2007 5:20 am 
Offline

Joined: Sun Jul 22, 2007 11:06 am
Posts: 259
JPack wrote:
This answers a lot of questions: http://www.dailytech.com/article.aspx?newsid=8221

Quote:
To his credit, the image on DT and every other publication was canned -- AMD gave it to media. They did not let media take their own pictures of the demo systems. Nor would they disclose what speeds the demo systems were running at.



Turns out the demo wasn't even running at 3 GHz. Now we know why it had a conventional heatsink and everything else seemed normal.


This is a flat out lie. Kris is 100% wrong here, and since I haven't talked to him about it, I can't say if it is on purpose or he simply botched it badly.

1) I was there.
2) The images on the Inq, both of them, are indeed different from every other publication. Look at the hands and the angles.
3) I took those pictures personally.
4) The systems were indeed running at 3.0, I personally played with both of them.
5) They let you do anything you wanted to on the systems short of taking benchmarks.


Kris's post is really odd. I wonder what prompted him to make up untrue things like that. He was not there, and I did not see a representative from his site there (but there very well could have been, I don't know).

This means (assuming didn't have someone there) Kris did not know what the press was told on site, not could he have known what I was told.

I would _SO_ love to hear the explanation for this one.

-Charlie


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 5:28 am 
Offline

Joined: Sun Jul 22, 2007 11:06 am
Posts: 259
Anonymous wrote:
IdRatherNotSayEither wrote:
Don't be stupid. It was running at 3 ghz, period. Running every application you could throw at 3 ghz.


And you know this how? I take it you were standing there as an AMD representative snapped the 3GHz picture of the demo system screen and handed it to the press, right?

If not, when exactly was the picture taken? Is there anyway of looking at the metadata to find out some more information about this picture?

Unless someone can verify that the image was taken from that system at that time all AMD proved is they can get one Phenom processor to boot into Windows using unknown cooling at unknown voltage with unknown stability.

The 2900XT UVD debacle is a recent reminder that AMD is adept at obscuring information and misleading the press.


I was standing there while the AMD representative (Jon Carvill) took the pictures. In fact, he borrowed _MY_ camera to do so, check the metadata.

It was taken between 8 and 8:30am PST last thursday at the annex building to the left of AMDs main lobby.

I can verify everything that Mr. Not Say says, and there were two machines there running at 3.0Ghz. I can also debunk all the FUD Kubicki is spewing over this one. I also know the cooling, the voltage, and a lot more info about it.

Basically, go away troll.

-Charlie


Top
 Profile  
 
 Post subject: Since you mention it.....
PostPosted: Tue Jul 31, 2007 5:46 am 
Offline

Joined: Sun Jul 22, 2007 11:06 am
Posts: 259
IdRatherNotSay wrote:
Anonymous wrote:
[
And you know this how? I take it you were standing there as an AMD representative snapped the 3GHz picture of the demo system screen and handed it to the press, right?


And maybe the machine those people got to see wasnt really connected to those displays either, did anyone check the wiring ? The real machine could have been in the basement somewhere with exotic or LN cooling applied to it.. Maybe it was an Intel quad core ? :) Maybe thats why they wherent allowed to take pictures, someone might see it.. ?

I guess AMD would think just like me: believe whatever you want. Not being allowed to takie pictures or run any app other than the one being demoed is an old industry practice in similar situations.


Well, since you brought the topic up, I did check the wiring, and had to move quite a bit of it when I took pictures of the rear of the machine. I can personally confirm that the keyboard/mouse/monitors shown were connected to the machine in question.

As for the pictures not being allowed, complete bull, I would be more than happy to send you some of the ones I took, front, rear, case open, case closed, and ones showing the monitor running the demos too.

-Charlie


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 10:07 am 
Offline

Joined: Wed Jun 27, 2007 1:45 pm
Posts: 59
Location: Germany
Groo's final Story & Pictures on this: http://uk.theinquirer.net/?article=41342
Apperently, he was doped by AMD - confess, Groo! ;-)


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 10:50 am 
Offline

Joined: Tue Jul 31, 2007 10:43 am
Posts: 23
Thanks Groo for strictly debunking the FUD that Intel special agent Kubicki is spreading over dailybullshit.
But, finances at AMD must be _very_ bad if they can't outfit their reps with digicams ...
BTW information about the voltage(s) would be very interesting!


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 11:19 am 
Offline

Joined: Tue Jul 31, 2007 11:04 am
Posts: 2
Good one Groo. Looking around there seems to be an awful amount of negative marketing in action where AMD's next platform is concerned. I wonder if AMD's competitor has a special department assigned purely to monitor forums in various high visibility tech sites.
As a platform how would this fare against the current best at the moment?
You may not have been allowed to run benchmarks specifically but did you get a sense of speed or more importantly power left in reserve?


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 1:28 pm 
Offline

Joined: Tue Jul 31, 2007 1:25 pm
Posts: 285
AMD's chips don't generally overclock much from the top bin, so the production process is probably close to 3ghz now. Given that they generally increase by ~50% as the process is run in, it looks like a 4.5ghz Barcelona will be facing 4.5ghz Penryn next year.

Which chip will have higher IPC (and therefore better performance)?


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 1:44 pm 
Offline

Joined: Wed Jun 27, 2007 1:38 pm
Posts: 668
If the "production process" would have been close to 3GHz now, AMD would be releasing "close to 3GHz" parts instead of 2GHz parts in the August...


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 2:32 pm 
Offline

Joined: Tue Jul 31, 2007 1:25 pm
Posts: 285
Consider Opteron.

All initial samples were limited to 800mhz (and didn't overclock).

By volume launch it was 1.8ghz

11 months later it was at 2.4ghz

All on the same 130nm process.
http://users.erols.com/chare/786.htm


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 3:23 pm 
Offline

Joined: Wed Jun 27, 2007 1:38 pm
Posts: 668
In this case Barcelona will be launched in August at 2GHz. If the ramp will be as slow as the Opteron ramp, then we should expect 2.6-2.7GHz parts in July 2008.

However, I think/hope Barcelona's ramp will be a bit faster, but I don't expect 3GHz parts anytime soon.


Last edited by jack on Tue Jul 31, 2007 7:17 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 4:41 pm 
Offline

Joined: Tue Jul 24, 2007 10:12 pm
Posts: 69
Quote:
AMD's chips don't generally overclock much from the top bin, so the production process is probably close to 3ghz now. Given that they generally increase by ~50% as the process is run in, it looks like a 4.5ghz Barcelona will be facing 4.5ghz Penryn next year.


I doubt that the production process is yielding at 3GHz. I would take that to mean that most of their chips are clocking at 3 GHz. I would expect these to be high binned parts, at least; it has been claimed that they are not cherry picked. I would say that to qualify as cherry picked, they would have to be at least 1 part in a 1000 or many more. They generally don't sell that many high-end parts anyway, so the high-end parts can yield at a low rate. I wouldn't expect any 3 GHz barcelona based Opterons soon. I don't think they will need the high clock speed for the HPC market or workstations. For servers (mostly branchy integer code) they may not be able to match C2, but performance isn't the only consideration. They may have a power advantage due to the on-die memory controller and the native quad-core (less IO power) and the use of DDR2 instead of FB-DIMM.

For Phenom branded chips, they will want the higher clock speed, but many phenom chips will be only dual core, which should yield at higher clock speeds. For a quad-core chip to yield at 3 GHz, than all 4 cores must be able to run at that speed; this does make the demo quite impressive though. Anyway, any chips comming off the production line right now will not make it to market for a while.

As for next year, who knows. AMD has been saying that they will have 45 nm comming relatively quickly, which may be true. Process research will be going on in parallel, so just because 65 nm is late does not mean that 45 nm will also be late. I don't think we will be seeing huge clock speed advances as we did in the past. These are not super pipelined chips like the P4. We will be seing more modest clock speed gains, but a lot more cores; 8 core/die at 45 nm. I am not really expecting AMD or Intel to hit 4 GHz on 65 nm, but I am not sure what is on the roadmaps.


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 31, 2007 4:46 pm 
Offline

Joined: Tue Jul 31, 2007 4:36 pm
Posts: 11
jack wrote:
In this case Barcelona will be launched in August at 2GHz. If the ramp will be as slow as the Opteron ramp, then we should expect 2.6-2.7GHz parts in July 2008.

However, I think/home Barcelona's ramp will be a bit faster, but I don't expect 3GHz parts anytime soon.

Yes, but when AMD launched their first Opteron, before that, they doesn't have any CPU faster than 2.25 Ghz. For less than one year the overtake clockspeed of their fastest clocked and far less complex Thoroughbred B core, with C0 Hammer, and after that with CG they are reached 2.6 Ghz.

Today we can buy 3.0 or 3.2 Ghz Athlon 64 at 90nm. Do you think that the AMD with 65nm process can't reach their own 90nm speeds ?
I don't think so, K10 dual core will run 3.2 ghz at 65nm for sure, next year, and quad will be clocked at 3.0 ghz.

I think that the AMD will launch 2.6 + Ghz CPU in 2008.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 91 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6, 7  Next

All times are UTC + 1 hour


Who is online

Users browsing this forum: No registered users and 0 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
suspicion-preferred