Opteron video encoding benches

[Brian Ellis]

Ray

No company is going to speculate on such an iffy situation. MS will do a truly minimal development to start with and will go on to a full-blown 64 bit solution only when it deems a return possible on the massive investment. MS and risk-taking are contradictory terms.

I don't believe you will be wowed: the MS development is based on W2k, NOT XP, and, as I say, it will be minimal. I have not heard of their developing any 64 bit apps, such as Office, either (not that Office would benefit from it: you won't be able to type any faster, just because you have a 64 bit OS).

And you may be right, AMD may not survive, but I'm sure they have contingency plans available to minimise the risk if 64 bit does go pear-shaped.

Finally, just ask yourself: if 64 bit were that attractive for mainstream PCs, why do you think Intel are not jumping on the bandwagon? The last I heard is that they have no plans to go that way before 2010 (other than for servers). It is not that they do not have the technology. If they see the A64 taking off, I'm sure the 'Hexium' could be launched within 2 or 3 months, probably already ready. Hardware is so much easier to design than OSs.

[bitz4brainz]

I don't know if this interests any of you (or if you are old enough to remember these companies), but I think there are some interesting parallels to the Digital Equipment and Data General late-70s transition to 32 bit minicomputers. The DG side was documented in the book "Soul of a New Machine" by Tracy Kidder.

First, there is the obvious David vs Goliath matchup. DEC was 9 times the size of DG. DG was always the scrappy underdog.

More importantly, there was the question of how much of a "nod" each company would give to backward compatibility with their legacy 16-bit systems.

By 1979, DEC had already come out with the 32-bit VAX and it was extremely popular with its customers. The VAX had a 16-bit "mode", but it was fundamentally a new, 32-bit machine.

DG was late to the party. They attempted two, starkly different approaches. The "official" successor machine (Fountainhead) was going to be a complete break from the legacy. In fact it used architectural concepts that were very far ahead of their time. The group that worked on it were thought to be DG's "best and brightest".

At the same time another group quietly set to work on an "upgrade to 32-bits" (Eagle) that could run both 16-bit and 32=bit software at the same time. In fact you could mix 16-bit and 32-bit instructions in the same program, while using 32-bit memory addressing overall. Like the Athlon 64, the Eagle's architecture was a logical extension of the earlier architecture. (But to the 32-bit purists, it was a kludge.)

The Fountainhead fell behind and never came to market. The Eagle was designed and built in about 18 months as a "skunkworks project". It became DG's 32-bit successor to their 16-bit line and was lauded by its customers for making the transition smoother than it might have been. DG's approach impressed some big new customers, like the US Forest Service, and the company's sales shot up.

Unfortunately, DG's success didn't last very long. The Intel revolution took out both companies, DG first.

The book is a great read, and won a Pulitzer prize. It's a case study in Machiavellian corporate politics, and an inside look at life as a bithead in a high-tech company. I recommend reading the book, but here is an article in Wired:

http://www.wired.com/wired/archive/8.12/soul.html

I guess my underlying point is that a major change in architectures is difficult for customers as well as the companies that build the CPUs, and that a pragmatic, transitional approach can help mitigate the pain. AMD's approach (an extension of the i86) seems to me to be like DG's. Intel's approach seems much like DEC's. And I would be inclined to bet on the underdog.

[Frank Marshall]

There's not too much new going on here. Wasn't Digital's Alpha chip was a 64-bit processor? Sounded so exciting at the time. Intel crushed 'em! SGI's hardware/software environment has been 64-bit for ages but they are becoming increasingly marginalised as well. I think it all comes back to the availability of application software for the hardware platform, "software leads hardware" goes the mantra. I'm worried AMD are putting the cart before the horse on this one.

[UtwigMU]

Well, it's mainly about ability to adress more than 4GB memory w/o loosing performance due to translations that occur with Xeon's 36-bit approach.

Ohterwise Hammer runs 32-bit code even better than Athlon.

I think it will become natural Athlon replacement and most people won't use 64-bitness. Except for Workstation and low cost (relatively) blade servers.

Just like most of people didn't run 32-bit applications on 386 when it came out in '87.

Intel crosslicenced x86-64 and it's rumored to be in Prescott but disabled (similar as Hyperthreading was disabled in sub 3.0GHz P4's). IIRC Tejas will support x86-64.

Now look at it this way: 32-bit CPU's became available in mid 80's, however it took untill late 90's for OS's to transit fully to 32-bit.

I think arround 2015 most people will run 64-bit computers and hardware.

[Brian Ellis]

"I think arround 2015 most people will run 64-bit computers and hardware."

I think Intel would agree with you, which is why they are not rushing on the bandwagon. AMD's rush is a last-chance gamble on survival.

"most people won't use 64-bitness."

Then why buy it and be forced to use a botched OS?

[Ray Austin]

Originally posted by Brian Ellis
Ray

No company is going to speculate on such an iffy situation. MS will do a truly minimal development to start with and will go on to a full-blown 64 bit solution only when it deems a return possible on the massive investment. MS and risk-taking are contradictory terms.

Sure they will, tech companies do it all the time... remember the Iridium thing I corrected you on, I know it's a bit different than this, but that was Xtremely speculative and look how that turned out. M$FT is a risk taker, but only if they know the money is there first... why did they join the gaming craze with Xbox. I think they are still losing money on that. So that's risk isn't it? It is M$FT right?

I don't believe you will be wowed: the MS development is based on W2k, NOT XP, and, as I say, it will be minimal. I have not heard of their developing any 64 bit apps, such as Office, either (not that Office would benefit from it: you won't be able to type any faster, just because you have a 64 bit OS).

well you don't need a 2000ghz 32bit machine to type either. That is a moot point. The fasest implementation for 64bit will be video games, no different than 32bit... and I'm hoping Adobe releases 64bit code for premiere, AE, photoshop etc.

And you may be right, AMD may not survive, but I'm sure they have contingency plans available to minimise the risk if 64 bit does go pear-shaped.

I think Amd will survive, it's in their best interest to be as aggressive as they can, that is their only hope at this point... I just love it when Amd pushes the envelope and I would be the first one in line for a 64/32bit machine.

Finally, just ask yourself: if 64 bit were that attractive for mainstream PCs, why do you think Intel are not jumping on the bandwagon? The last I heard is that they have no plans to go that way before 2010 (other than for servers). It is not that they do not have the technology. If they see the A64 taking off, I'm sure the 'Hexium' could be launched within 2 or 3 months, probably already ready. Hardware is so much easier to design than OSs.

Ok, I can answer that in one word... ITANIUM Naturally they aren't going to bring a cheap 64bit desktop chip to market when they can charge thousands of dollars for one in the commercial space for servers. That's why Amd is a thorn in Intels side at the moment. If anyone is good at losing money and sticking around it's AMD. They have been in that position many times... only Amd knows how thorny their path is at the moment. I think the next school season is very critical for them. Hopefully the A64 is released without any more delays... they should do fine..... I hope!

[Ray Austin]

Originally posted by Frank Marshall
There's not too much new going on here. Wasn't Digital's Alpha chip was a 64-bit processor? Sounded so exciting at the time. Intel crushed 'em! SGI's hardware/software environment has been 64-bit for ages but they are becoming increasingly marginalised as well. I think it all comes back to the availability of application software for the hardware platform, "software leads hardware" goes the mantra. I'm worried AMD are putting the cart before the horse on this one.

Yes true, it's a shame it wasn't exploited. But different time and place and 64bit then is different than 64 bit now. For one the speeds of the CPU are much higher which will funtionally make 64 bit more feasible.

No, Amd is fine. For one the alpha chip couldn't due 32 bit. But the new A64 can, remember different time, different chip. The game is very much different now. Also Amd is not the only one going to 64bit... apple is moving there too. Coincidence???

Could there be an Apple/ Amd connection??? Something to think about anyway...

[Ray Austin]

Originally posted by Brian Ellis
"I think arround 2015 most people will run 64-bit computers and hardware."

I think Intel would agree with you, which is why they are not rushing on the bandwagon. AMD's rush is a last-chance gamble on survival.

"most people won't use 64-bitness."

Then why buy it and be forced to use a botched OS?

Oh yeah and what we use now isn't botched??? hahahahahahaha

You can be funny when you want to be.... hahahahahahahahaha