AMD issues STOP SHIP order on Barcelona

Printable View

Show 100 post(s) from this thread on one page

12-03-2007, 11:09 PM
terrace215

Quote:

Originally Posted by Zytek_Fan

Obviously what's happening is B3 stepping! :D

B3 is not until mid-to-late Q1, officially, from AMD.

Samples will be sooner, obviously.

Let's hope they don't end up needing a B4.
12-03-2007, 11:22 PM
Zytek_Fan

Quote:

Originally Posted by terrace215

B3 is not until mid-to-late Q1, officially, from AMD.

Samples will be sooner, obviously.

Let's hope they don't end up needing a B4.

Well, I've already lost all hope of me getting a Phenom that rocks before March, so Yorkfield is what will power my new rig :)
12-03-2007, 11:41 PM
NightCrawler™

I've just been reading the other thread.... now this... [censored]
12-03-2007, 11:50 PM
Epsilon84

Quote:

Originally Posted by LOE

Am I the only one who thinks this is good news? Yeah AMD will miss another 2-3 months, but if thats true... that means K10 is not weak, but broken.

If indeed the fix results in low clockspeed and 10-20% performance hit, resolvin the bug will turn K10 into a solid competitor, even on the desktop front....

In server, a 3GHz K10 with 10-20% higher IPC will wipe the floor with anything intel can offer as good as K8 did with pentium 4

fudzilla already posted there are bug free samples of K10... I hope thats true and B3 is finally ready and fully working

I think you are a little confused. The fix does NOT increase IPC by 10 - 20%, it merely fixes stability at the expense of 10 - 20% performance.

Therefore, performance will be 10 - 20% LOWER than the review scores, but with guaranteed stability.

So no, this is not good news at all. In fact, it is horrible news. A '100% stable' K10 with the fix would be reduced to K8 levels of IPC... or even lower!
12-03-2007, 11:53 PM
dinos22

Quote:

Originally Posted by LOE

Am I the only one who thinks this is good news?

yeap you are the only one :shrug:
12-04-2007, 12:46 AM
Epsilon84

Quote:

Originally Posted by LOE

So reviews are without the fix? But retail K10 uses the fix?
Is there someone who can confirm that?

Obviously, since the BIOS fix is still 'in the works', meaning unreleased. Therefore, how can reviews be using the fix, when it didn't even exist?

And no, retail K10s currently DON'T have the fix. It takes a BIOS update, which is yet to be released. Once released, you can have the option of the fix, but with a performance penalty.

Logic is your friend. :up:
12-04-2007, 12:56 AM
Epsilon84

Quote:

Originally Posted by LOE

This is big news, I wonder how come there is no site posting it?

Tech Report is... check the other thread: http://www.xtremesystems.org/forums/...d.php?t=168066
12-04-2007, 01:06 AM
Epsilon84

Quote:

Originally Posted by LOE

yeah, I read the first page of the thread... it is quite obvious TR reports it... but no one else

I mean news like this should be all over the internet by now

Give it time... it's hardly in the best interests of AMD for this to be spreading like wildfire anyway. Keeping it low key, if possible, would be the best result for AMD at this point.
12-04-2007, 01:10 AM
KTE

Quote:

Originally Posted by savantu

AFAICT the TLB bug doesn't diminish performance on current low clocked parts , the fix applied does that.

It does on mine which is a low clocked part. Mine is slower than everyone elses I've seen at slower speeds.

Quote:

In other words , K10 won't increase IPC over what the 2.2/2.3GHz parts tested have shown.

This means nothing here. The bug affects performance. Meaning 2.42GHz beats 2.7GHz in Super Pi 1M and Cinebench 10. So it definitely makes a big difference in performance.
12-04-2007, 01:13 AM
Epsilon84

Quote:

Originally Posted by KTE

It does on mine which is a low clocked part. Mine is slower than everyone elses I've seen at slower speeds.This means nothing here. The bug affects performance. Meaning 2.42GHz beats 2.7GHz in Super Pi 1M and Cinebench 10. So it definitely makes a big difference in performance.

The bug affects STABILTY. Perhaps performance is affected slightly as well, in the case of L3 cache misses, but the updated BIOS doesn't fix that. It actually turns off parts of the TLB that are causing stability problems.

Again, the fix improves stability, not performance. How hard is that to understand?
12-04-2007, 03:16 AM
iddqd

Quote:

Originally Posted by Omastar

Unlike AMD, Intel has the immense R&D and raw capital to put up in order to implement an IMC correctly.

Believe it or not, modulating has its benefits over integrating. A modular memory controller (such as one located on a north bridge chip) allows for more platform flexibility without the need to have redundant logic that tends to ramp R&D and production costs.

Besides, with their fast (and large) L2 caches, Intel is less reliant on RAM than AMD.
12-04-2007, 03:28 AM
informal

Quote:

Originally Posted by Epsilon84

The bug affects STABILTY. Perhaps performance is affected slightly as well, in the case of L3 cache misses, but the updated BIOS doesn't fix that. It actually turns off parts of the TLB that are causing stability problems.

Again, the fix improves stability, not performance. How hard is that to understand?

Maybe because he has the chip and has tested it and seen the lower clocked part beat up the 200MHz higher clock??:rolleyes:
12-04-2007, 03:58 AM
Dainas

Quote:

Originally Posted by iddqd

A modular memory controller (such as one located on a north bridge chip) allows for more platform flexibility without the need to have redundant logic that tends to ramp R&D and production costs.

Yep, 939 is a great example of worse case scenario... The AM2+ AM3 stopgap has shown that AMD must have felt as if it was dumping some of its 939 users into Intels lap by that. And now it looks as if Intel will be doing the same with Nehalem and its IMC new socket 771 deal. Will AMD have some worth flocking to at that point though as 939 users saw in Core 2? :shrug: One can only hope as it sure as heck is not enticing in mass right now.
12-04-2007, 04:37 AM
happychappy

Quote:

Originally Posted by LOE

Am I the only one who thinks this is good news? Yeah AMD will miss another 2-3 months, but if thats true... that means K10 is not weak, but broken.

If indeed the fix results in low clockspeed and 10-20% performance hit, resolvin the bug will turn K10 into a solid competitor, even on the desktop front....

In server, a 3GHz K10 with 10-20% higher IPC will wipe the floor with anything intel can offer as good as K8 did with pentium 4

fudzilla already posted there are bug free samples of K10... I hope thats true and B3 is finally ready and fully working

Things must be bad when you're relying on Fudzilla for some good news:ROTF:
12-04-2007, 07:07 AM
EvilBellyLint

Quote:

Originally Posted by MR_SmartAss

Nope, you are not the only one. I think this is a good news too. The AMD fanpois are suffering. :rofl:

Dear Mr. SmartAss,

Please try and contribute something, anything, constructive once in a while?

Flaunting your self esteem issues is becoming rather... embarrassing.

your pal,

EBL :up:
12-04-2007, 07:19 AM
EvilBellyLint

Quote:

Originally Posted by Tibu

I propose an Official XS Betting Pool on estimated date of AMD bankruptcy...

I was thinking the other day, how the overwhelming negative sentiment
in light of all this bad AMD press is starting to smack of a good contrarian
indicator. But when you begin to see posts about betting pools on a
company's demise, and the vast majority of the general public agrees,
the truly smart money will take the opposite side of that bet. ;)

EBL
12-04-2007, 08:42 AM
n91htmare

I love AMD, it's the fanbois that ruin it for me.
12-04-2007, 09:01 AM
Donnie27

Quote:

Originally Posted by EvilBellyLint

Intel will not drop prices due to this. They have shareholders to
think of, and it's not like Intel is chasing AMD in market-share...

EBL

Let me try again then. Intel will both Drop Prices on some products, build some new products to be cheaper from the start and then Raise prices on others. NOTHING will be across-the-board=P Intel's product line-up diverse enough to both lower and raise prices and make even more money. I think some folks approach Intel in an AMD kind of way. AMD's Inventory is very limited in Volume and Range. Intel is building low-end products for a Profit, not a loss or to hold onto market share. I was just saying their stockholders are happy and been that way since Conroe launched.

TO; Jumpingjack

Intel built up Goodwill when they had too. This started with them announcing 800MHz NorthwoodC when they didn't have too. They hurt sells of their own NW-A when they did this. Webmasters were given Hotscott and full excess to all of its info, good and bad. T-R even pointed this out. Info was sent out about Pentium-M long before anyone had one. Folks got their hands on them many months before they went on sale, same goes for all of its follow-on derivative up to and including Conroe.

I'd always said if AMD has something they'd show it and was promptly flamed LOL! Of course, not by you.
12-04-2007, 10:43 AM
Jacky

I dont want to open another thread on AMD's miserable execution lately so I'll ask now and here:

What does AMD's sudden TDP increase mean? (and a thread on RWT about the issue) Different binning to get better yields? Or is the combination of 65nm node + Barc. arch that bad atm?
They say the ACP stays the same (even though some people argue ACP is just marketing BS and would prefer to stay with TDP).

If ACP 75W results in TDP 115W (55W ACP-> 79W TDP; 105W -> 137W), are all the B2 stepping Barcelonas @2ghz already 115W? wow, if true. or are they just relaxing the TDP envelopes for future products to get better yields and allow 2.5ghz+ Barcelonas which would exceed 120W TDP.
As far as I know there have been rumours about 2.6ghz phenom/barc @ prescott-like 120W+.
12-04-2007, 10:54 AM
Shintai

ACP is just another bad marketing attempt by AMD because they cant compete with Intel. And they dont want to have a preshott on them after their arrogant statements a few years back. They did more than the BS you could expect from a company.

And seriously...the guy updating the errata got vacation? In a company with over 10000 employees. And they try tell me they depend on 1 single guy? Maybe if it was a 10 man company...

Show 100 post(s) from this thread on one page