Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

No 3GHz Soon and No G5 Laptop

sorry, didn't mean to start another tangent

thatwendigo said:
It's a bit misleading...
Click to expand...

But the current P4 and Xeon chips also have the same defensive capability.

I only mentioned this thermal protection in the context of the story about IBM's reluctance to run the PPC970 at 2.0 GHz in the JS20.

The IBM engineers probably felt that there were common failure modes (dead chassis fans, blocked airflow, high room temperature) that could not be survived with the chip clocked at 2.0 GHz.

I once visited a Compaq test lab where they had some prototype PIII Xeon workstations undergoing thermal testing. They literally had the motherboard and power supply in an "oven" where it was being "baked" while running exercisers.

Something caught my eye - the SCSI cable snaked out of the oven to a disk on the workbench. I asked about that - the response was:
"We know the disk will burn up at those temperatures - but we're not testing the disk."

I have no doubt but that IBM has baked the JS20 as well....
 
AidenShaw said:
But the current P4 and Xeon chips also have the same defensive capability.

I only mentioned this thermal protection in the context of the story about IBM's reluctance to run the PPC970 at 2.0 GHz in the JS20.

The IBM engineers probably felt that there were common failure modes (dead chassis fans, blocked airflow, high room temperature) that could not be survived with the chip clocked at 2.0 GHz..
Click to expand...

It's amazing that people feel the need to come up with bizarre hypothese to explain something that's already well known.

The IBM system in question is a blade server. Blade servers have a very high density of chips and removing the heat is always a problem. In the PC world, blade servers rarely have the latest chip - and NEVER if each blade has dual processors.

You just can't run a blade server at as high a clock speed as a conventional server. It just can't be done. There are too many chips generating too much heat in too small an area. You're forced to use a chip below the top end. That's true of every blade vendor.

Why generate bizarre speculation to try to get around that simple explanation?
 
with every post, you can be so wrong

jragosta said:
In the PC world, blade servers rarely have the latest chip - and NEVER if each blade has dual processors.

You just can't run a blade server at as high a clock speed as a conventional server. It just can't be done. There are too many chips generating too much heat in too small an area. You're forced to use a chip below the top end. That's true of every blade vendor.
Click to expand...


Hmmm, is that why IBM's HS20 is using the latest 3.2 GHz 2MiB cache Xeons in their dual processor blades? (Note - the HS20 is the same form factor, and uses the same chassis, as the JS20 PPC blades. You can even use both HS20 and JS20 blades in the same chassis at the same time.)

Is that why IBM's HS40 is using four of the latest 3.0 GHz 4MiB cache Xeon MPs?

http://www-1.ibm.com/servers/eserver/bladecenter/index.html


Is that why HP's BL20p is using the latest 3.2 GHz 2MiB cache Xeons in their dual processor blades?

Is that why HP's BL40p is using four of the latest 3.0 GHz 4MiB cache Xeon MPs?

http://h18004.www1.hp.com/products/servers/proliant-bl/p-class/index.html


You are just too full of yourself - you just made statements about x86 blades, presenting them as absolute fact.

However, in just a few seconds of surfing the major blademakers sites, your claims can be shown to be completely false.

I can see why macsrus decided to stop trying to discuss things with you....
 
AidenShaw said:
Hmmm, is that why IBM's HS20 is using the latest 3.2 GHz 2MiB cache Xeons in their dual processor blades? (Note - the HS20 is the same form factor, and uses the same chassis, as the JS20 PPC blades. You can even use both HS20 and JS20 blades in the same chassis at the same time.)

Is that why IBM's HS40 is using four of the latest 3.0 GHz 4MiB cache Xeon MPs?

http://www-1.ibm.com/servers/eserver/bladecenter/index.html


Is that why HP's BL20p is using the latest 3.2 GHz 2MiB cache Xeons in their dual processor blades?

Is that why HP's BL40p is using four of the latest 3.0 GHz 4MiB cache Xeon MPs?

http://h18004.www1.hp.com/products/servers/proliant-bl/p-class/index.html


You are just too full of yourself - I can see why macsrus decided to stop trying....
Click to expand...


Hey, thanks for proving my point.

None of those is the fastest chip available - the P4 Extreme (3.4 GHz, 800 MHz frontside bus).

Not to mention, of course, that when I go to the URLs you provided, the fastest chip shown for either the IBM or HP servers is the 3.0 GHz Xeon - which isn't even the fastest available Xeon (you can easily get 3.2 GHz).

So, in spite of your ranting, all you've done is prove that I'm right. BOTH URLs you provided support my position. Neither IBM nor HP uses the fastest Xeon (and certainly not the fastest P4) in their blade servers.

I love it when you make a fool of yourself.
 
what was that remark about shifting goal posts?

jragosta said:
I love it when you make a fool of yourself.
Click to expand...


Look again, the 3.2 GHz Xeon (DP) is on the dual CPU blades - which is the latest, fastest available dual CPU chip. Period.

The 3.0 GHz Xeon MP is on the quad blades, and it's the latest, fastest available chip with quad capabilities. Period.

The 3.4EE is a different chip - it's an over-clocked Xeon DP at its heart, but it's in the 478 pin package that does not support dual processing. It's a one-of for the gamer market, not a server chip. And of course, they're not putting single-processor-capable chips in dual processor machines, even Cybil Fawlty would understand that.

So, the blades do support the *latest* chips, as you claimed they did not. To quibble about the 3.4EE vs the Xeon server line is nonsense.

And, of course, your earlier comment was along the lines of "sure IBM underclocks the 2.0 - you have to for a blade". IBM and HP aren't underclocking the fastest Xeons for their blades.

Do, please go back to those URLs and find the 3.2GHz Xeon DP blades. It would be a healthy exercise in reading comprehension for you.
 
AidenShaw said:
Look again, the 3.2 GHz Xeon (DP) is on the dual CPU blades - which is the latest, fastest available dual CPU chip. Period.

The 3.0 GHz Xeon MP is on the quad blades, and it's the latest, fastest available chip with quad capabilities. Period.

The 3.4EE is a different chip - it's an over-clocked Xeon DP at its heart, but it's in the 478 pin package that does not support dual processing. It's a one-of for the gamer market, not a server chip. And of course, they're not putting single-processor-capable chips in dual processor machines, even Cybil Fawlty would understand that.

So, the blades do support the *latest* chips, as you claimed they did not. To quibble about the 3.4EE vs the Xeon server line is nonsense.

And, of course, your earlier comment was along the lines of "sure IBM underclocks the 2.0 - you have to for a blade". IBM and HP aren't underclocking the fastest Xeons for their blades.

Do, please go back to those URLs and find the 3.2GHz Xeon DP blades. It would be a healthy exercise in reading comprehension for you.
Click to expand...

Oops. Sorry - I just looked at the quad from HP.

But you're STILL supporting my point. The dual HP uses the 3.2. The Quad uses the 3.0 (max). And there are still lots of 2.0 GHz CPUs in these servers.

That merely demonstrates what I've been saying all along. It's quite common both in the x86 world and in the G5 world to require slower chips when you have a large number of chips inside the box. Your own URLs support that.

I can't believe you're arguing such an obvious point.
 
jragosta said:
I can't believe you're arguing such an obvious point.
Click to expand...

And I can't believe that you're suggesting that they put in chips that have the wrong number of pins for the sockets!

Those same 3.2DP and 3.0MP chips are the ones that go into the 1U, 2U, 4U and 7U servers.

They are *not* putting slower chips in the blades vs. the larger servers (which actually have *more* chips in the box, albeit a larger box).

Your point about having to downrate the CPUs for the compact spaces inside blade servers is simply wrong. The same top speed chips are used in all form factors.

I'll remind you what you said earlier, just in case you forgot where the goal posts are:

jragosta said:
The IBM system in question is a blade server. Blade servers have a very high density of chips and removing the heat is always a problem. In the PC world, blade servers rarely have the latest chip - and NEVER if each blade has dual processors.

You just can't run a blade server at as high a clock speed as a conventional server. It just can't be done. There are too many chips generating too much heat in too small an area.
Click to expand...

So, they are *exactly* running the blade servers and the conventional servers at the same speed. You are wrong, pure and simple. Changing the question to be about the desktop EE chip is moving the goal posts. You are wrong about the servers.
 
jgrosta


macsirus originally said IBM simply didnt want to stand behind a 2.0 blade server as their own. NOT because a 1.6 would be over clocked, but becuase they predicted a great eal of heat related issues.

simple.

all he is saying is that they didnt want to make a 2.0 themselves. and if they havent...then ...that would be ...CORRECT. (note..i don['t care enough to check as to whether ANY 2.0 IBM blades are sold with those specific chips...
 
Alright. That's enough.

To make my money every day I am the Americas Product Manager for the IBM eServer BladeCenter.

IBM in *no way* underclocks, overclocks, or uses 'old processors' in their BladeCenter servers (The HS20 and HS40) or any server we sell.

The HS20 uses the Intel Xeon Processor. These processors come from Intel at their rated speeds of 2.8GHz, 3.06GHz, and 3.2GHz. The 3.2GHz comes in two flavors, a 1MB L2 cache and a 2MB L2 cache. These are the absolute fastest processors available from Intel that support dual processors. They access the front side bus at 533MHz

The HS40 uses The Intel Xeon Processor MP. This processor comes in three flavors. There is the 2.2GHz and 2.7GHz with 2MB L3 cache, and the 3.0GHz with 4MB of L3 cache. They access the front side bus at 400MHz. These are the fastest processors that support more than 2 in a system.

We have no problems removing the heat from said processors and do it very well. As a matter of fact Gartner just published a study that showed us to be one of the most efficient in dissapating said heat and using the least amount of power. We use a liquid core heat sink on the processors and then two dual blowers that are in the chassis. While I wouldn't want to sit next to one every day (they are a bit loud) it is quite efficient and the fans only run at most 40% of their rated capacity with the hottest, fastest processor in there (the 3.2GHz/2MB).

While I do not manage the JS20 (PowerPC 970) directly, we announced this product in November of 2003. At that time 1.6GHz was where we wanted to be. This in no way means we're done and don't want to 'stand behind' a 2.0GHz version of said product. While I cannot publicly comment without an NDA I believe I can say we're 'not done yet' with this product and will continue to do other things with it. Also the PowerPC 970 at 1.6GHz in our chassis consumes the same power/puts out about the same heat as a 2.8GHz processor from Intel. Heat/Power of the 2.0 was not the issue, when we announced it was.

While HP maybe my competitor, they too use the exact same processors in the Intel space, we just believe we do it better. :D I'm not sure if their heat sinks have a liquid center but they do put 9 fans on each four processor blades and 6 on each two processor blade (BL40p and BL20p) last I looked. They do not support a non-Intel processor in their blade server at this time.

While I am in no way representing IBM in any posting I ever make for fun and hobby here, I felt I should set the record straight.
 
Again for at least the 3rd or 4th time...
IBM decided not to use a CPU(PPC970 130nm or the First 970s made at 90nm )clocked faster than 1.6, in their JS20 blade centers....
because the original 970 and the first 90nm 970s (whether they were fxs are not im not positive.) Did not have thermal/power throttle control.

Therefore They felt they were not suitable for use at the higher speeds, in the blades because of potential heat related issues.

The 970s/970fxs that are being produced right now do have the power/thermal control and according to IBMs roadmap they will use 2.2 GHZ 3rd Quarter 2004
Also according to their roadmap for the CPUs in the Blades they will increase to 3GHZ second Quarter 2005 all the way to 6GHZ by 4th Quarter 2006.
at least thats the way I remember seeing it.

By the way... As I have said before the reason the first 970s did not have power/thermal control was because Apple said it wasnt important to them in the First series of this CPU.
Which as I have also previously stated is related to IBMs decision as to why they were not going to use them clocked faster than 1.6 in their first blades
 
Surreal said:
jgrosta


macsirus originally said IBM simply didnt want to stand behind a 2.0 blade server as their own. NOT because a 1.6 would be over clocked, but becuase they predicted a great eal of heat related issues.

simple.

all he is saying is that they didnt want to make a 2.0 themselves. and if they havent...then ...that would be ...CORRECT. (note..i don['t care enough to check as to whether ANY 2.0 IBM blades are sold with those specific chips...
Click to expand...

That's NOT what he said. Go back and read it. He specifically said that they were underclocking 2.0 GHz chips and selling them as 1.6. I disagreed with that - and pointed out how silly it would be. The IBM guy specifically said they do NOT do that. Macsrus was wrong. Period.
 
macsrus said:
Again for at least the 3rd or 4th time...
IBM decided not to use a CPU(PPC970 130nm or the First 970s made at 90nm )clocked faster than 1.6, in their JS20 blade centers....
because the original 970 and the first 90nm 970s (whether they were fxs are not im not positive.) Did not have thermal/power throttle control.

Therefore They felt they were not suitable for use at the higher speeds, in the blades because of potential heat related issues.
Click to expand...

The only problem is that this is the FIRST time you've said that.

Your original post was that IBM was underclocking 2.0 GHz chips at 1.6. I pointed out how silly that was and you kept insisting that it was true.

The IBM guy says you're wrong. Period.
 
Trekkie said:
To make my money every day I am the Americas Product Manager for the IBM eServer BladeCenter.
Click to expand...

Thanks for the info - the HS20 (I have a couple chassis) is a nice product.

A couple of questions, though:

1. When will the HS20 with 3.6GHz/800MHz Nocona CPUs be available?

2. Will IBM come out with a system to compete with HPaq's BL10e system (20 blades in a 3U chassis)? I currently have 8 of these (160 CPUs) and have 4 more on order (80 more CPUs) [I had some left over $$ this quarter]. They're really sweet for scale-out problems when you don't need absolute top end performance per thread - roughly 30 GHz of P4 power in a 3U chassis.

I'd love to see IBM compete in this space - twenty to twenty-four 2GHz Dothan CPUs in a 3U chassis would be nice!

Feel free to decline to answer publicly ;) !
 
You don't know the business, do you?

jragosta said:
Your original post was that IBM was underclocking 2.0 GHz chips at 1.6. I pointed out how silly that was and you kept insisting that it was true.

The IBM guy says you're wrong. Period.
Click to expand...


#1 - the IBM guy is the *Intel Blade* product manager. Do you think that he has spare cycles to stay up-to-date on minor technical details of all the other IBM products?

#2 - the IBM guy is a *product manager* - not an IBM engineer. In large tech organizations, the job of the product manager is to be the "Great Communicator" - he needs to tell engineering what the customer needs, and he has to describe to the customer what engineering has built. The product manager does not design or control the product, that's an engineering function. Engineering might have decided to use 2.0 chips in the 1.6 blade without telling product management about that minor detail.

While I have immense respect for the vital role of product management in a large company, if an end user pulls the heat sink off the chip and says that the chip is stamped "2.0" - I'll believe the end user.

And yes, I have personally pulled the heat sinks off my BladeCenter boards and swapped and shuffled the CPUs. (Mine are labeled 2.4GHz, though....)
 
AidenShaw said:
I'll believe the end user.
Click to expand...

There isnt any reason to believe anything I think or say....
I have spent sometime going over about 100 of Jagrosta's posts and he has never been wrong or misinformed about anything.....

Even IBM and Apple should hire him to straighten out their sells, engineering, manufacturing, and marketing departments.....

Their stockholders would benifit immensly.....

We should all bow humbly to his superior knowledge, wisdow, and intellect....
We should all stop posting and just wait on him to post.... so he can teach us about everything....

P.S. Im sure he is very goodlooking too
 
macsrus said:
There isnt any reason to believe anything I think or say....
I have spent sometime going over about 100 of Jagrosta's posts and he has never been wrong or misinformed about anything.....

Even IBM and Apple should hire him to straighten out their sells, engineering, manufacturing, and marketing departments.....

Their stockholders would benifit immensly.....

We should all bow humbly to his superior knowledge, wisdow, and intellect....
We should all stop posting and just wait on him to post.... so he can teach us about everything....

P.S. Im sure he is very goodlooking too
Click to expand...

Are you being sarcastic? ;) :p
 
AidenShaw said:
#1 - the IBM guy is the *Intel Blade* product manager. Do you think that he has spare cycles to stay up-to-date on minor technical details of all the other IBM products?

#2 - the IBM guy is a *product manager* - not an IBM engineer. In large tech organizations, the job of the product manager is to be the "Great Communicator" - he needs to tell engineering what the customer needs, and he has to describe to the customer what engineering has built. The product manager does not design or control the product, that's an engineering function. Engineering might have decided to use 2.0 chips in the 1.6 blade without telling product management about that minor detail.

While I have immense respect for the vital role of product management in a large company, if an end user pulls the heat sink off the chip and says that the chip is stamped "2.0" - I'll believe the end user.

And yes, I have personally pulled the heat sinks off my BladeCenter boards and swapped and shuffled the CPUs. (Mine are labeled 2.4GHz, though....)
Click to expand...


I see. So you know more about it than the IBM product manager?

He says that IBM doesn't use underclocked CPUs. You say they do.

Is there any plausible reason why anyone should believe you?
 
jragosta said:
I see. So you know more about it than the IBM product manager?
Click to expand...

Please reread (the reading comprehension problem again) the Trekkie post.

He is *not* the product manager for the PowerPC blade servers, but for the *Intel* blade servers. He admits that he doesn't know the all the details about the PPC product.


jragosta said:
He says that IBM doesn't use underclocked CPUs. You say they do.
Click to expand...

Once again, you have trouble understanding a post.

I did not say that IBM is using underclocked CPUs. I said that if a user pulls the heatsink and sees "2.0" stamped on the CPU chip in his system - that I'd believe that user over a product manager.

If the engineering manager for the PPC blades said that they weren't clocking 2.0 chips at 1.6, I'd certainly consider that as more authorative than a statement from a product manager for a different IBM product. Even then, if the chip is stamped "2.0 GHz" which are you going to believe?


jragosta said:
Is there any plausible reason why anyone should believe you?
Click to expand...

You should believe that it is *my opinion* that I'd trust what is stamped on the CPU of the shipping product, rather than a statement by someone in product management or marketing.

An engineering change like that can happen without the ECO being broadcast all the way through the engineering, product management and marketing chains.


Give up while you're behind....
 
AidenShaw said:
Please reread (the reading comprehension problem again) the Trekkie post.

He is *not* the product manager for the PowerPC blade servers, but for the *Intel* blade servers. He admits that he doesn't know the all the details about the PPC product.




Once again, you have trouble understanding a post.

I did not say that IBM is using underclocked CPUs. I said that if a user pulls the heatsink and sees "2.0" stamped on the CPU chip in his system - that I'd believe that user over a product manager.

If the engineering manager for the PPC blades said that they weren't clocking 2.0 chips at 1.6, I'd certainly consider that as more authorative than a statement from a product manager for a different IBM product. Even then, if the chip is stamped "2.0 GHz" which are you going to believe?




You should believe that it is *my opinion* that I'd trust what is stamped on the CPU of the shipping product, rather than a statement by someone in product management or marketing.

An engineering change like that can happen without the ECO being broadcast all the way through the engineering, product management and marketing chains.


Give up while you're behind....
Click to expand...

What we have are a few facts:

1. The IBM product manager for Intel blade servers says that IBM does NOT underclock systems.

2. You insist that they may be underclocking systems.

Yes, you are NOW saying 'if the chips says 2.0'. The fact is that you have no reason to believe that the chip would say 2.0. IBM's product manager for a related product line says that it does not. Common sense says it does not. Standard business practice says that it does not.

Who to believe?
 
jragosta said:
What we have are a few facts:

1. The IBM product manager for Intel blade servers says that IBM does NOT underclock systems.

2. You insist that they may be underclocking systems.

Yes, you are NOW saying 'if the chips says 2.0'. The fact is that you have no reason to believe that the chip would say 2.0. IBM's product manager for a related product line says that it does not. Common sense says it does not. Standard business practice says that it does not.

Who to believe?
Click to expand...


Your reading skills have failed you again - there's another fact that you need to deal with:


macsrus said:
[#693]
I have a question for U.. Do you own xserves? Do you own JS20 blade servers?

We do... I have seen the CPUs.... I am not speculating.

[#694]
AS I said in my last post I am not speculating on IBM using some 2.0s clocked at 1.6 in their JS20... we own a rack of them... (6 blade centers 14 blades ea 2 CPUs per blade)

I pulled the heat sinks on a blade they were 2.0s... I cant say all of them are, but the 2 I looked at were.
Click to expand...


Who to believe?

Is macsrus flat out lying when he says that he has seen "2.0" engraved on the actual CPUs in an actual JS20 blade that he has, or is the product manager of one IBM product line not completely tuned into what the engineering and manufacturing teams for a different IBM product line are doing?

Having worked in software and hardware development in a couple of the largest computer companies, it is my opinion that either one is possible. I lean towards believing macsrus, because I have no reason to believe that he would be lying (and I've seen product managers who were confused about minor technical details of their *own* products, let alone a different product line).
_______________________

I'll sign off as well, jragosta, until you come up with some fresh reasoning. This debate with you is turning into a Monty Python skit....

"I came here for an argument. An argument is not just contradiction.
Yes it is.
No it's not.
Yes it is!
No it's not!
..."
 
AidenShaw said:
Your reading skills have failed you again - there's another fact that you need to deal with:

Who to believe?

Is macsrus flat out lying when he says that he has seen "2.0" engraved on the actual CPUs in an actual JS20 blade that he has, or is the product manager of one IBM product line not completely tuned into what the engineering and manufacturing teams for a different IBM product line are doing?[/i]
Click to expand...

Actually I believe IBM's Product manager 100% on not underclocking or over clocking The INTEL CPUs in their HS20 and HS40 product lines
Also I Believe 100% IBM never overclocks a CPU used in ANY System they have ever sold.

The real question is whether or not they have ever used a CPU that marked at a higher rated speed as a slower part....
I know for a fact that is true as I personally have a JS20 blade with 2.0 CPUs on it.....

Something else to think about is.... I got my first blade center before Apple started delivering Xserve G5s..... It is possible that the CPUs in the blade Center we have are 130nm.... also ill have to check some of my newer blades and see what speed is stamped on those CPUs.

I have grown completly tired of this whole discussion about them.
And after I check some Monday ill retire from posting in this thread....
Of course as Ive said before it doesnt really matter what I say because some people choose to call me a liar anyway.
 
AidenShaw said:
Your reading skills have failed you again - there's another fact that you need to deal with:





Who to believe?

Is macsrus flat out lying when he says that he has seen "2.0" engraved on the actual CPUs in an actual JS20 blade that he has, or is the product manager of one IBM product line not completely tuned into what the engineering and manufacturing teams for a different IBM product line are doing?

Having worked in software and hardware development in a couple of the largest computer companies, it is my opinion that either one is possible. I lean towards believing macsrus, because I have no reason to believe that he would be lying (and I've seen product managers who were confused about minor technical details of their *own* products, let alone a different product line).
_______________________

I'll sign off as well, jragosta, until you come up with some fresh reasoning. This debate with you is turning into a Monty Python skit....

"I came here for an argument. An argument is not just contradiction.
Yes it is.
No it's not.
Yes it is!
No it's not!
..."
Click to expand...


Macsrus never claimed to have seen at 2.0 GHz chip on an IBM system sold as 1.6. Read his posts again.

So you're basically saying that we should believe you over the IBM product manager because of some fictitious information from your sock puppet.

OK.

The interesting thing is that I don't need to come up with any new information. IBM has supported my position. You're the one living in a fantasy world.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back