6 months of delay due to someone else's IP not being up to spec

[ElFenix]

Although, publicly, ATI representatives wouldn't lay blame on exactly were the issue existed, quietly some will point out that when the issue was eventually traced it had occurred not in any of ATI's logic cells, but instead in a piece of "off-the-shelf" third party IP whose 90nm library was not correct. One the issue was actually traced, after nearly 6 months of attacking numerous points where they felt the problems could have occurred, it took them less than an hour to resolve in the design, requiring only a contact and metal change, and once back from the fab with the fix in place stable, yield-able clockspeeds jumped in the order of 160MHz.

that really blows

 

[Hacp]

Originally posted by: ElFenix

Although, publicly, ATI representatives wouldn't lay blame on exactly were the issue existed, quietly some will point out that when the issue was eventually traced it had occurred not in any of ATI's logic cells, but instead in a piece of "off-the-shelf" third party IP whose 90nm library was not correct. One the issue was actually traced, after nearly 6 months of attacking numerous points where they felt the problems could have occurred, it took them less than an hour to resolve in the design, requiring only a contact and metal change, and once back from the fab with the fix in place stable, yield-able clockspeeds jumped in the order of 160MHz.

that really blows

Ya I read it and wish I had a time machine. I would hav ethen offered the solution to ATI, and they could have given me like what 4/5 of the tapeout costs .

 

[lifeguard1999]

According to the Inq, the guy who found the problem got a nice bonus. Kudos to her/him!

 

[ArchAngel777]

Originally posted by: lifeguard1999
According to the Inq, the guy who found the problem got a nice bonus. Kudos to her/him!

I am pretty skeptical of this in the first place. It is easy to blame someone else for the problem so that the blame shifts to another company. Regardless, if it took them 6 months to fix this problem, I question the technical ability of the company.

Just my opinion... But I am not sure I believe this.

 

[sxr7171]

Originally posted by: ElFenix

Although, publicly, ATI representatives wouldn't lay blame on exactly were the issue existed, quietly some will point out that when the issue was eventually traced it had occurred not in any of ATI's logic cells, but instead in a piece of "off-the-shelf" third party IP whose 90nm library was not correct. One the issue was actually traced, after nearly 6 months of attacking numerous points where they felt the problems could have occurred, it took them less than an hour to resolve in the design, requiring only a contact and metal change, and once back from the fab with the fix in place stable, yield-able clockspeeds jumped in the order of 160MHz.

that really blows

Well better luck to them next time. They're going to need it.

 

[ElFenix]

Originally posted by: ArchAngel777

Originally posted by: lifeguard1999
According to the Inq, the guy who found the problem got a nice bonus. Kudos to her/him!

I am pretty skeptical of this in the first place. It is easy to blame someone else for the problem so that the blame shifts to another company. Regardless, if it took them 6 months to fix this problem, I question the technical ability of the company.

Just my opinion... But I am not sure I believe this.

well, they said the engineers were being very candid about what was going wrong. engineers tend to be that way, rather than the marketing folks who would do something like what you've suggested. at least, from my experience. i imagine tracking down random-seeming ghosts in 325 million transistors could be hard. they said it only took an hour to fix, it was a bitch to find, though.

 

[Cruise51]

Not sure what kind of a problem would avoid detection for 6 months, but then again, I'm not exactly an expert on the matter. Would be nice to be the guy who found it, $$$:thumbsup:.

 

[Gstanfor]

Already posted a reply in another thread, but I think it deserves a repeat post.

Originally posted by: Beyond3D article
According to public reports ATI noticed that as late as July, issues occurred that prevented the R520 core being clocked close to its target speeds, which is consistent with leakage issues. Curiously, the issue was not consistent across all their 90nm products - ATI had already delivered Xenos to Microsoft using the same 90nm process R520 does, and other derivatives of the R520 line suffered the same issue (RV530) but others did not (RV515) - the fact R520 and RV530 share the same memory bus, while RV515 and Xenos have different memory busses is not likely to be coincidental in this case. ATI were open about talking about the issue they faced bringing up R520, sometimes describing the issue in such detail that only Electronic Engineers are likely to understand, however their primary issue when trying to track it down was that it wasn't a consistent failure - it was almost random in its appearance, causing boards to fail in different cases at different times, the only consistent element being that it occurs at high clockspeeds. Although, publicly, ATI representatives wouldn't lay blame on exactly were the issue existed, quietly some will point out that when the issue was eventually traced it had occurred not in any of ATI's logic cells, but instead in a piece of "off-the-shelf" third party IP whose 90nm library was not correct. One the issue was actually traced, after nearly 6 months of attacking numerous points where they felt the problems could have occurred, it took them less than an hour to resolve in the design, requiring only a contact and metal change, and once back from the fab with the fix in place stable, yield-able clockspeeds jumped in the order of 160MHz.

What's the bet that the third party IP in question was the much vaunted Fast-14 math logic that ATi supporters felt would make such a difference to ATi's designs? It certainly seems to have made a difference, though perhaps not the one ATi and its supporters were hoping for... Meanwhile, nVidia has been quietly benefitting from Arithmatica's CellMath librarys for close on 2.5 years now... :laugh:

 

[Hacp]

Originally posted by: ArchAngel777

Originally posted by: lifeguard1999
According to the Inq, the guy who found the problem got a nice bonus. Kudos to her/him!

I am pretty skeptical of this in the first place. It is easy to blame someone else for the problem so that the blame shifts to another company. Regardless, if it took them 6 months to fix this problem, I question the technical ability of the company.

Just my opinion... But I am not sure I believe this.

Well, It says that ATI doesn't want to admit it, so ATI isn't blaming anyone. Its IQ blaming people.

 

[Genx87]

Originally posted by: ElFenix

Although, publicly, ATI representatives wouldn't lay blame on exactly were the issue existed, quietly some will point out that when the issue was eventually traced it had occurred not in any of ATI's logic cells, but instead in a piece of "off-the-shelf" third party IP whose 90nm library was not correct. One the issue was actually traced, after nearly 6 months of attacking numerous points where they felt the problems could have occurred, it took them less than an hour to resolve in the design, requiring only a contact and metal change, and once back from the fab with the fix in place stable, yield-able clockspeeds jumped in the order of 160MHz.

that really blows

Have to wonder the capability of ATI if it took them 6 months to figure this out.

 

[ArchAngel777]

Originally posted by: Hacp

Originally posted by: ArchAngel777

Originally posted by: lifeguard1999
According to the Inq, the guy who found the problem got a nice bonus. Kudos to her/him!

I am pretty skeptical of this in the first place. It is easy to blame someone else for the problem so that the blame shifts to another company. Regardless, if it took them 6 months to fix this problem, I question the technical ability of the company.

Just my opinion... But I am not sure I believe this.

Well, It says that ATI doesn't want to admit it, so ATI isn't blaming anyone. Its IQ blaming people.

Again, I am skeptical of this. It isn't hard for ATI to fake this. It could be the truth, but I have doubts on this.

 

[rbV5]

I don't find it hard to believe at all. You have to wonder how hard they would have been looking for it if it hadn't directly affected core speed. NV40 shipped with a broken video processor, and obviously they knew it, and shipped it anyway weighing the delayed launch of another NV40 tape-out vs the lack of the missing feature. While I would have prefered nvid1a would have been up-front (even till this day), it was a call they made, and it worked out OK for them evenif it alienated a small part of their userbase. Obviously this defect would have had far reaching consequences compared to Nvidias broken video processor, and ATI looks to have made the right choice even if it did really set them back.