first tri-cores in the wild

[SexyK]

Originally posted by: keysplayr2003

Originally posted by: taltamir
they did bfdd, they were unable to complete even a single photoshop benchmark... now it wasn't crashing on loading photoshop. But their benchmark involved something that would trigger the crash. (photoshop has batch configurations... which I use often...)

If I remember correctly, there was a bug with Photoshop (forget what version) years ago that only happened when using a PIII 1.13GHz Tualatin? It occured in Photoshop in only one operation, and that was "Rotate" with a certain degree (45 or 90). That was due to the CPU errata and confirmed by Intel. So nobody is perfect.

The difference is, when Intel discovered that bug they halted shipments for 4+ months if I recall correctly and didn't foist a clearly bugged chip on unsuspecting consumers. Not really the same situation.

 

[VirtualLarry]

Originally posted by: CTho9305
Huh? The 1.13GHz Tualatins failed on plenty of stuff. I remember Tom's Hardware couldn't get through a linux kernel compile - both Tom's and HardOCP had multiple issues with those chips. I can't seem to find the original articles any more though .

That was the original PIII 1Ghz or 1.13Ghz - the coppermine. Not the tualatin.

 

[Arkaign]

Originally posted by: VirtualLarry

Originally posted by: CTho9305
Huh? The 1.13GHz Tualatins failed on plenty of stuff. I remember Tom's Hardware couldn't get through a linux kernel compile - both Tom's and HardOCP had multiple issues with those chips. I can't seem to find the original articles any more though .

That was the original PIII 1Ghz or 1.13Ghz - the coppermine. Not the tualatin.

Exactly. The Tualatin P3 was a .13mu chip that had an IPC somewhere between Athlon and Athlon XP. There were 256 and 512k cache versions, they had an IHS, and they ran very cool.

 

[magreen]

Originally posted by: myocardia

Originally posted by: Viditor

Originally posted by: myocardia
Yet as of today, there's only 17 motherboards that will POST with a Phenom, and most of those seem to have the newest chipsets: http://products.amd.com/en-us/RecommendedMBFilter.aspx

What??? Just because a mobo isn't on the recommended list doesn't mean it won't work with Phenom...those are just the mobos that AMD has certified to now (there are many more that aren't on the list but still work fine).

Which ones, specifically? Tom's looked, but couldn't seem to find them. Notice the name of the article, btw?

Yeah, it's a pretty damning article.
OTOH, knowing THG they probably stuck the chip in a toaster and declared it incompatible.

 

[myocardia]

Originally posted by: VirtualLarry

Originally posted by: CTho9305
Huh? The 1.13GHz Tualatins failed on plenty of stuff. I remember Tom's Hardware couldn't get through a linux kernel compile - both Tom's and HardOCP had multiple issues with those chips. I can't seem to find the original articles any more though .

That was the original PIII 1Ghz or 1.13Ghz - the coppermine. Not the tualatin.

That's correct, it was the 1.13 Ghz Coppermine cores, which were the second PIII cores. The first PIII cores were the Katmai, which were 250nm. And IIRC, there wasn't a bug with the 1.13 Ghz Coppermines, they were just overheating, when used with the crappy stock Intel heatsink. Or am I just remembering wrong?

 

[Arkaign]

http://www.tomshardware.com/20...ntium_iii_1/index.html

 

[Modelworks]

I think this speaks volumes for Amd.
If they were doing well they would have halted all production, corrected the issue and then started selling the cpu.

It seems Amd is too far in the hole to be able to go that long without some income from the chips. Unfortunate , the market needs competition.

I know when I was with GE, that if a semiconductor had a problem and it was decided to ship as is, that they simply cut the connection with the pin that feature went to on the chip. No way to re-enable it

They also didn't reveal what problem the original chip had, instead the whole chip was renamed and you would just see on the datasheet pins marked NC for no connection.

 

[myocardia]

Originally posted by: Arkaign
http://www.tomshardware.com/20...ntium_iii_1/index.html

Shouldn't this be something more along the lines of You do infact have a bad memory? That was a good read. I don't remember it at all, though. Fortunately for me (I guess), they say that memory is the first to go. Great, I wonder what's next?:Q It had better not be continence, or I'm gonna be pissed, pun intended.

 

[Duvie]

No way AMD comes up with a bios fix that takes a 10% hit in performance for a 1 in a million chance of error. I call BS on that one....If it was as rare as the fanbots say and only when running vmware they likely would not have needed a fix at all for the average consumer that buys desktop PCs....Not for a 10% PENALTY. Heck most of the average users I know would not even blink if their PC got an occasional program error or crash. They would likely assume it is software and MicroSuck was to blame. Another clear sign to me that the 1 in a million claim was pulled out of someones arse.

It is bigger then the fanbots and AMD wants to publicly admit.


As I see it....There isn't a reason for me to even look cross-eyed at AMDs offerings at this point. I will likely get a new 45nm quad-core in June....Otherwise my oc'd QX6700 still will outperform the best AMD has to offer similarly Oc'd on air.


I wonder if the cache intensive app like running multiple instances of Folding At Homes new SMP program would error? And then whether it would crash out with the error or send false data back. I think distributed computing ppl should stay far away from phenoms until they can get a bug free version. I have noticed since getting my C2 Duals that F@H hits the cache hard and cache makes a huge difference in many of the workunits...Hence why the AMDs suck clock for clock doing F@H.

 

[jiffylube1024]

Originally posted by: CTho9305

Originally posted by: keysplayr2003

Originally posted by: taltamir
they did bfdd, they were unable to complete even a single photoshop benchmark... now it wasn't crashing on loading photoshop. But their benchmark involved something that would trigger the crash. (photoshop has batch configurations... which I use often...)

If I remember correctly, there was a bug with Photoshop (forget what version) years ago that only happened when using a PIII 1.13GHz Tualatin? It occured in Photoshop in only one operation, and that was "Rotate" with a certain degree (45 or 90). That was due to the CPU errata and confirmed by Intel. So nobody is perfect.

Huh? The 1.13GHz Tualatins failed on plenty of stuff. I remember Tom's Hardware couldn't get through a linux kernel compile - both Tom's and HardOCP had multiple issues with those chips. I can't seem to find the original articles any more though .

No that was the old Coppermine 1.13 GHz that was pulled from the shelves before it was ever released that THG found all of the problems with.

The Tualatin was the 130nm version of the P3, and Intel released Tualatins from 1.1-1.4 GHz.

Tualatin came out just after the P4 was launched. It was the only P3 version to include a metal heatspreader over the die.

 

[harpoon84]

Btw, Phenom tri and dual core benches:

http://www.pcgameshardware.de/?article_id=622354

http://www.erenumerique.fr/tes...l_ve_-art-1804-11.html

 

[sandorski]

Originally posted by: jiffylube1024

Originally posted by: CTho9305

Originally posted by: keysplayr2003

Originally posted by: taltamir
they did bfdd, they were unable to complete even a single photoshop benchmark... now it wasn't crashing on loading photoshop. But their benchmark involved something that would trigger the crash. (photoshop has batch configurations... which I use often...)

If I remember correctly, there was a bug with Photoshop (forget what version) years ago that only happened when using a PIII 1.13GHz Tualatin? It occured in Photoshop in only one operation, and that was "Rotate" with a certain degree (45 or 90). That was due to the CPU errata and confirmed by Intel. So nobody is perfect.

Huh? The 1.13GHz Tualatins failed on plenty of stuff. I remember Tom's Hardware couldn't get through a linux kernel compile - both Tom's and HardOCP had multiple issues with those chips. I can't seem to find the original articles any more though .

No that was the old Coppermine 1.13 GHz that was pulled from the shelves before it was ever released that THG found all of the problems with.

The Tualatin was the 130nm version of the P3, and Intel released Tualatins from 1.1-1.4 GHz.

Tualatin came out just after the P4 was launched. It was the only P3 version to include a metal heatspreader over the die.

Hmm, as I recall it, the Photoshop Bug was with all Coppermines and not just the 1ghz+. The 1ghz+ issue was simply one of a CPU being pushed beyond stability, resulting in all kinds of problems, but first detected due to a failed Linux Compile/Install at Tom's.