Could this mean G5 PowerBooks announced in January? It will be interesting to see what happens.
What does the increase in the number of G5 processors IBM pumps out in a day have to do with designing one that can be squeezed into the PowerBook form factor?
What exactly are they saying though? It never says they getting better yields. It never says they are going to producing the 40% more from fishkill either.
I hate to be pessimistic, but Business has a way of putting an overly positive spin on what is actually happening. I wish it would have said they are seeing 40% better yields on 90nm G5 PPC processors.
Not to say that this is not good news that 40% more processors will be coming from somewhere, but I think you see what I'm saying.
What does the increase in the number of G5 processors IBM pumps out in a day have to do with designing one that can be squeezed into the PowerBook form factor?
Less power leakage, fewer defects limiting chip operations. Your low power chips are actually usually the best of the batch.
To those who see this as "IBM should've done this months ago", do you have any idea how chip manufacturing works? They ran into issues going to 90nm and were working at fixing the issue. Why would IBM want to get a, perhaps, 50% yield and produce a bunch of useless wafers when they were having problems getting things right? Seriously, you can't believe that!
Now that yields are up, so is the volume. What part of this is hard to understand?
You know, IBM is not just producing them for Apple alone....they use them too in their JS20 blade servers. People, come on....
What exactly are they saying though? It never says they getting better yields....
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
For a while now, I've been wondering if the 2.3GHz used @ Virginia Tech just might be preproduction runs of the 970 using SOI? Maybe these are some of the chips that Chipworks refered to when saying they were expecting production qualifying run samples of the 970 using SOI. I have no basis for this, but it seems that the xServers would need lower power cpu's than desktops, maybe?? Plus the fact that the 2.3GHz hasn't appeared @ the Apple Store yet.
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
For a while now, I've been wondering if the 2.3GHz used @ Virginia Tech just might be preproduction runs of the 970 using SOI? Maybe these are some of the chips that Chipworks refered to when saying they were expecting production qualifying run samples of the 970 using SOI. I have no basis for this, but it seems that the xServers would need lower power cpu's than desktops, maybe?? Plus the fact that the 2.3GHz hasn't appeared @ the Apple Store yet.
I really doubt that -- they are probably just hand-picked cream-of-the-crop processors. They would have become 2.5s but the VT needed them cooler than that so they were down clocked.
The IBM announcement doesn't say how they will accomplish this increase in production. Note that it says production and not yields which implies that they are either devoting more of their 300mm capacity to the G5 (now that the yields are good enough to make it profitable), or they have spent the money to increase their 300mm capacity (now that the yields are good enough to make it profitable). Either way this isn't something they could have done before because of the yield problems. If you are losing X million per day per production line, why would you add production lines? They had enough running to deliver some product to Apple and to work out the kinks in the system, but only when the yields rise to a certain level can they afford to ramp production. Now it seems that yields have improved to that point, which is a very good sign. The 2.5 yields are still iffy, but that will likely have to await the next revision of the 970 family.
To those who see this as "IBM should've done this months ago", do you have any idea how chip manufacturing works? They ran into issues going to 90nm and were working at fixing the issue. Why would IBM want to get a, perhaps, 50% yield and produce a bunch of useless wafers when they were having problems getting things right? Seriously, you can't believe that!
Now that yields are up, so is the volume. What part of this is hard to understand?
You know, IBM is not just producing them for Apple alone....they use them too in their JS20 blade servers. People, come on....
Duuuude, stop interjecting reality into this discussion. We are having fun apart from the real world We all know that making a large quantity of small chips is very easy. After all, Motorola could do it, so why not IBM?
Stupid IBM! Out to destroy Apple! Too little, too late! You twits! Why you curse Apple product? When I get headless G5 for $300?
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
For a while now, I've been wondering if the 2.3GHz used @ Virginia Tech just might be preproduction runs of the 970 using SOI? Maybe these are some of the chips that Chipworks refered to when saying they were expecting production qualifying run samples of the 970 using SOI. I have no basis for this, but it seems that the xServers would need lower power cpu's than desktops, maybe?? Plus the fact that the 2.3GHz hasn't appeared @ the Apple Store yet.
What IBM said was -according to AI
Quote:
Originally noted by AI
"IBM said it will significantly increase its microprocessor production during the current quarter"
What your saying that it implies it never actually says. What it does say is the will be producing more processors using the 300-mm wafers.
This could mean any # of things. They may be slowing, or stopping production on some older less used fabs, and devoting more 300-mm wafer stations (terminology isn't my strong point here) that will end up producing about 40% more G5 processors.
If they got specific, and noted that yields on the existing 300-mm wafers were up by 40% I would see this as improvement to the fab process, but this press release has no indication of actual progress with the manufacturing process pertaining to individual 300-mm wafer yields.
[edit] - didn't see your post programmer, but good points - all of them.
Seems quite clear to me. He wants to know when the next PowerPC CPU after the 970FX is coming. We've had the 970, and the 970FX so far... when's the next CPU comin'.
Seems quite clear to me. He wants to know when the next PowerPC CPU after the 970FX is coming. We've had the 970, and the 970FX so far... when's the next CPU comin'.
What does the increase in the number of G5 processors IBM pumps out in a day have to do with designing one that can be squeezed into the PowerBook form factor?
I wasn't talking about the design. That a totally different issue. I'm saying that perhaps IBM is going to produce more G5's because Apple may be planning to release the PowerBook with a G5 processor come January. For all we know, Apple may have already addressed the heat/battery life issues. You don't know, I don't know, nobody knows, except Apple engineers.
not AI - Silicon Strategies unless AI reported inaccurately.
Quote:
Originally posted by onlooker
What your saying that it implies it never actually says. What it does say is the will be producing more processors using the 300-mm wafers.
Quote:
Originally posted by onlooker
What exactly are they saying though? It never says they getting better yields.
What it does say is," However, the company it was enthusiastic over progress being made by IBM to increase G5 yields." Unless of course AI isn't reporting from the Silicon Strategies article.
I did say, your right-refering to increased wafer production, but you painted a very negative picture about yields, when in fact IBM has stated that yields have been improving and only the 2.5GHz yields are yet to come up to expectations.
Quote:
Originally posted by Programmer
I really doubt that -- they are probably just hand-picked cream-of-the-crop processors. They would have become 2.5s but the VT needed them cooler than that so they were down clocked.
Oh well, hope springs eternal. At least Chipworks was expecting samples a qualifying production run of a 970fx using SOI soon and that was reported over a month ago. I wonder what "soon" means to Chipworks?: a month, 3 months, a year? time will tell.
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
I'm not trying to spark a big argument here, but like you said the 2.5 could take longer. Why? Read what Programmer says first.
Quote:
Originally posted by Programmer
Note that it says production and not yields which implies that they are either devoting more of their 300mm capacity to the G5 (now that the yields are good enough to make it profitable), or they have spent the money to increase their 300mm capacity (now that the yields are good enough to make it profitable). Either way this isn't something they could have done before because of the yield problems. If you are losing X million per day per production line, why would you add production lines? They had enough running to deliver some product to Apple and to work out the kinks in the system, but only when the yields rise to a certain level can they afford to ramp production.
IBM is aware that Apple must have processors now, Like yesterday. If just Production, and not yields per 300-mm wafer is increasing.
It seems only accurate that the 2.5Ghz would be least profitable to hand pick off the wafers off of because increasing production would turn less of a loss adding an additional 40% of production lines if one line was loosing considerably more that the others. Like only 3% of the chips were making it.
So adding 40% with two lines that are turning out better yields will turn over more product faster.
That is why I think the 2.5 is going to take longer. I think they are going to turn out enough of the other two, and stockpile them, and hopefully by then they can get at least 10% (or however much) instead of 3% yields, and turn the whole line to the 2.5, and start cranking out as many as possible using like 80% of the entire line for 2.5GHz versions.
Comments
Originally posted by s_sarinana
Could this mean G5 PowerBooks announced in January? It will be interesting to see what happens.
What does the increase in the number of G5 processors IBM pumps out in a day have to do with designing one that can be squeezed into the PowerBook form factor?
I hate to be pessimistic, but Business has a way of putting an overly positive spin on what is actually happening. I wish it would have said they are seeing 40% better yields on 90nm G5 PPC processors.
Not to say that this is not good news that 40% more processors will be coming from somewhere, but I think you see what I'm saying.
They seek it there
Those Macies seek it everywhere
Is it in heaven
Or is it in hell
That damned elusive 3GHz G5!
;-)
Originally posted by LudwigVan
What does the increase in the number of G5 processors IBM pumps out in a day have to do with designing one that can be squeezed into the PowerBook form factor?
Less power leakage, fewer defects limiting chip operations. Your low power chips are actually usually the best of the batch.
Originally posted by debuysserk
sure, this is great news, but they should've done this a couple of months ago.
Yes, but if this actually happens, I see a good future.
Now that yields are up, so is the volume. What part of this is hard to understand?
You know, IBM is not just producing them for Apple alone....they use them too in their JS20 blade servers. People, come on....
Originally posted by onlooker
What exactly are they saying though? It never says they getting better yields....
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
For a while now, I've been wondering if the 2.3GHz used @ Virginia Tech just might be preproduction runs of the 970 using SOI? Maybe these are some of the chips that Chipworks refered to when saying they were expecting production qualifying run samples of the 970 using SOI. I have no basis for this, but it seems that the xServers would need lower power cpu's than desktops, maybe?? Plus the fact that the 2.3GHz hasn't appeared @ the Apple Store yet.
Originally posted by rickag
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
For a while now, I've been wondering if the 2.3GHz used @ Virginia Tech just might be preproduction runs of the 970 using SOI? Maybe these are some of the chips that Chipworks refered to when saying they were expecting production qualifying run samples of the 970 using SOI. I have no basis for this, but it seems that the xServers would need lower power cpu's than desktops, maybe?? Plus the fact that the 2.3GHz hasn't appeared @ the Apple Store yet.
I really doubt that -- they are probably just hand-picked cream-of-the-crop processors. They would have become 2.5s but the VT needed them cooler than that so they were down clocked.
The IBM announcement doesn't say how they will accomplish this increase in production. Note that it says production and not yields which implies that they are either devoting more of their 300mm capacity to the G5 (now that the yields are good enough to make it profitable), or they have spent the money to increase their 300mm capacity (now that the yields are good enough to make it profitable). Either way this isn't something they could have done before because of the yield problems. If you are losing X million per day per production line, why would you add production lines? They had enough running to deliver some product to Apple and to work out the kinks in the system, but only when the yields rise to a certain level can they afford to ramp production. Now it seems that yields have improved to that point, which is a very good sign. The 2.5 yields are still iffy, but that will likely have to await the next revision of the 970 family.
Does anyone know when they'll release a new PowerPC? I''m now considering one of those.
Originally posted by cubedcompanies
well i'd think they'd have to to meet demand of iMac.
Does anyone know when they'll release a new PowerPC? I''m now considering one of those.
What on earth did you just ask?
PowerBook? PowerMac?
Originally posted by Rhumgod
To those who see this as "IBM should've done this months ago", do you have any idea how chip manufacturing works? They ran into issues going to 90nm and were working at fixing the issue. Why would IBM want to get a, perhaps, 50% yield and produce a bunch of useless wafers when they were having problems getting things right? Seriously, you can't believe that!
Now that yields are up, so is the volume. What part of this is hard to understand?
You know, IBM is not just producing them for Apple alone....they use them too in their JS20 blade servers. People, come on....
Duuuude, stop interjecting reality into this discussion. We are having fun apart from the real world
Stupid IBM! Out to destroy Apple! Too little, too late! You twits! Why you curse Apple product? When I get headless G5 for $300?
Originally posted by rickag
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
For a while now, I've been wondering if the 2.3GHz used @ Virginia Tech just might be preproduction runs of the 970 using SOI? Maybe these are some of the chips that Chipworks refered to when saying they were expecting production qualifying run samples of the 970 using SOI. I have no basis for this, but it seems that the xServers would need lower power cpu's than desktops, maybe?? Plus the fact that the 2.3GHz hasn't appeared @ the Apple Store yet.
What IBM said was -according to AI
Originally noted by AI
"IBM said it will significantly increase its microprocessor production during the current quarter"
What your saying that it implies it never actually says. What it does say is the will be producing more processors using the 300-mm wafers.
This could mean any # of things. They may be slowing, or stopping production on some older less used fabs, and devoting more 300-mm wafer stations (terminology isn't my strong point here) that will end up producing about 40% more G5 processors.
If they got specific, and noted that yields on the existing 300-mm wafers were up by 40% I would see this as improvement to the fab process, but this press release has no indication of actual progress with the manufacturing process pertaining to individual 300-mm wafer yields.
[edit] - didn't see your post programmer, but good points - all of them.
Originally posted by tak1108
What on earth did you just ask?
PowerBook? PowerMac?
Seems quite clear to me. He wants to know when the next PowerPC CPU after the 970FX is coming. We've had the 970, and the 970FX so far... when's the next CPU comin'.
Originally posted by Leonard
Seems quite clear to me. He wants to know when the next PowerPC CPU after the 970FX is coming. We've had the 970, and the 970FX so far... when's the next CPU comin'.
970GX with 2MB L2 cache.
Originally posted by LudwigVan
What does the increase in the number of G5 processors IBM pumps out in a day have to do with designing one that can be squeezed into the PowerBook form factor?
I wasn't talking about the design. That a totally different issue. I'm saying that perhaps IBM is going to produce more G5's because Apple may be planning to release the PowerBook with a G5 processor come January. For all we know, Apple may have already addressed the heat/battery life issues. You don't know, I don't know, nobody knows, except Apple engineers.
Originally posted by onlooker
What IBM said was -according to AI
not AI - Silicon Strategies unless AI reported inaccurately.
Originally posted by onlooker
What your saying that it implies it never actually says. What it does say is the will be producing more processors using the 300-mm wafers.
Originally posted by onlooker
What exactly are they saying though? It never says they getting better yields.
What it does say is," However, the company it was enthusiastic over progress being made by IBM to increase G5 yields." Unless of course AI isn't reporting from the Silicon Strategies article.
I did say, your right-refering to increased wafer production, but you painted a very negative picture about yields, when in fact IBM has stated that yields have been improving and only the 2.5GHz yields are yet to come up to expectations.
Originally posted by Programmer
I really doubt that -- they are probably just hand-picked cream-of-the-crop processors. They would have become 2.5s but the VT needed them cooler than that so they were down clocked.
Oh well, hope springs eternal. At least Chipworks was expecting samples a qualifying production run of a 970fx using SOI soon and that was reported over a month ago. I wonder what "soon" means to Chipworks?: a month, 3 months, a year? time will tell.
Originally posted by rickag
Your right, but it does say,"Yields of IBM's 2.5GHz G5 processor could take slightly longer to improve, Apple warned.". Longer than what? Which kind of implies they are getting improved yields on the 2.0, 1.8 and possibly the 2.3gHz(?).
I'm not trying to spark a big argument here, but like you said the 2.5 could take longer. Why? Read what Programmer says first.
Originally posted by Programmer
Note that it says production and not yields which implies that they are either devoting more of their 300mm capacity to the G5 (now that the yields are good enough to make it profitable), or they have spent the money to increase their 300mm capacity (now that the yields are good enough to make it profitable). Either way this isn't something they could have done before because of the yield problems. If you are losing X million per day per production line, why would you add production lines? They had enough running to deliver some product to Apple and to work out the kinks in the system, but only when the yields rise to a certain level can they afford to ramp production.
IBM is aware that Apple must have processors now, Like yesterday. If just Production, and not yields per 300-mm wafer is increasing.
It seems only accurate that the 2.5Ghz would be least profitable to hand pick off the wafers off of because increasing production would turn less of a loss adding an additional 40% of production lines if one line was loosing considerably more that the others. Like only 3% of the chips were making it.
So adding 40% with two lines that are turning out better yields will turn over more product faster.
That is why I think the 2.5 is going to take longer. I think they are going to turn out enough of the other two, and stockpile them, and hopefully by then they can get at least 10% (or however much) instead of 3% yields, and turn the whole line to the 2.5, and start cranking out as many as possible using like 80% of the entire line for 2.5GHz versions.
It's just a theory.