Amiga.org
Amiga computer related discussion => Amiga Hardware Issues and discussion => Topic started by: on March 13, 2003, 12:00:49 AM
-
MacGuardians (German) report from CeBit that IBM's PowerPC 970 will debut at up to 1.8GHz as originally expected. (IBM's Microprocessor Forum presentation in October 2002 indicated initial speeds for the PowerPC 970 ranging from 1.4GHz - 1.8GHz.)
The 2.5GHz models described in an IBM press release more recently are reportedly for the subsequent generation of 970's, but will apparently utilize the 0.13 Micron Process, contrary to ZDNet's report.
http://www.macrumors.com/pages/2003/03/20030312094607.shtml
-
PPC970 looks interesting now that the top speed will be 2.5ghz and not 1.8... at 2.5ghz it might hold the speed crown for a time.
-
It'll be ridiculously expensive I would imagine
-
yeah probably like 800$ per-chip or so... remember what the current top-end G4's cost...and just tack on a little extra.
It'll be good for servers and low/middle end workstations though...looking at it objectively it will probably cost alot and eventually fall clearly behind X86-64... but 64bit PowerPC at 2.2ghz will be something to see. If it holds the speed crown or at least is competative I'll probably buy a Mac ;)
-
reality check, high end is always pricey:
http://www.theinquirer.net/?article=8223 (http://www.theinquirer.net/?article=8223)
-
mips_proc wrote:
yeah probably like 800$ per-chip or so... remember what the current top-end G4's cost...and just tack on a little extra.
It'll be good for servers and low/middle end workstations though...looking at it objectively it will probably cost alot and eventually fall clearly behind X86-64... but 64bit PowerPC at 2.2ghz will be something to see. If it holds the speed crown or at least is competative I'll probably buy a Mac ;)
Does the Itanium (intels 64bit solution) even exceed 2ghz? Besides its not backwards compatible to the slightly more marketable Clawhammer? AMD 64bit.
Another thing, I am pretty sure that the x86 64bit CPUs are almost entirely different architecture and doesn't hold the same blistering Ghz speeds as their 32bit brethren. So it may seem PPC 970 might have the edge. Whether it holds onto it, is another quesion altogether.
-
What's interesting to note about the 970 is that it's a cut-down high-clock version of the POWER4 series of CPUs, which has already proven itself in battle over a number of years. Both the 64 bit X86 alternatives have so far failed to reach the market (unless you count 900MHz Itanium 2 chips for $2700 as reaching the market).
However, prediction is never accurate, even more so when it comes to predicting the future ;-)
So for now we must continue to take anything that's presented as "fact" with a grain of salt. If the 970 can indeed hit the market in reasonable clock ranges at a reasonable price it will most probably be something of a success. But how much really depends on how Intel/AMD go from here. Do they continue to produce faster and faster 32 bit chips, or start migrating people over to 64 bit? Impossible to say. But I really expect Intel to reach something of an upper limit in clock frequencies soon.
Edit: I would just like to comment that I'm not saying POWER4 ain't been as expensive as Itanium 1 and 2. I meant to make the point that the 970 is a more marketable version of a battle-proofed chip, whereas Itanium 2 is still only for HIGH END :-)
-
Another thing, I am pretty sure that the x86 64bit CPUs are almost entirely different architecture
Define "entirely different architecture" for K8 vs K7 families.
and doesn't hold the same blistering Ghz speeds as their 32bit brethren.
Refer to http://amiga.org/forums/showthread.php?t=43216
-
Hmm, the "Unverified source" story? :-)
-
Both the 64 bit X86 alternatives have so far failed to reach the market
That would be false in regards to AMD's Opteron. Refer to AMD's "beachhead" programs (some sort of “early bird” program).
Define “hitting the market" (i.e. you seems to apply different “hitting the market" definitions on different products).
IBM's PPC970 64bit CPU is not the only one to attend the CeBIT show.
Refer to http://www.theinquirer.net/?article=8274
Summary
- Several mass-producing X86 motherboard manufactures will be showing their X86-64 motherboards products i.e. these include Tyan, MSI, Gigabyte and Asus.
- MS's Windows 64bit (beta build) edition will be also on show.
- AMD Athlon 64 laptops.
(unless you count 900MHz Itanium 2 chips for $2700 as reaching the market).
Note that IA-64 was based HP's PA-RISC. Secondly, what was the price for Power 4 CPUs again?
Careful with your claims or assertions.
-
You didn't read the comment I edited in, did you? I feel I properly answered BOTH of those questions BEFORE you asked them.
-
Note that IA-64 was based HP's PA-RISC. Secondly, what was the price for Power 4 CPUs again?
Weren't CBM thinking of making next gen Amiga's using PA-RISC processors?
-
the itanium is essentially in 'beta' right now... as much as it's bieng marketed it isnt bieng marketed at the 'desktop' its in a higher-class then the desktop.. it's priced out of our hands for a reason... the application and OS support isnt there... I've got a friend with an Itanium and it's not very fast nor does the WinXP for it hold a candle in the stability department to the X86 deskto-class chips.
X86-64 will probably be marketed right at the desktop from the get-go it's a desktop chip... it isnt here yet...and nor will there be X86-Win2K/XP for it initially when it arrives that takes advantage of it bieng 64bit... aswell as no X86-64 apps initially.
I personally think Intel will go X86-64 in the end just because of market pressure.
If the PPC970 comes out...and it kicks ass...and I could get a mac with two of them for 3000$... even if their 1.8's?... I'd go for it...if they where competative... I think they will be
as of right now macintrash's are pathetic in my eyes... their non-competative and they offer lousey benchmarks even compared to single X86 machines..
yes commodore's next generation Amiga was going to use PA-Risc wich is direct relative to Itanium...more like an inbred uncle then a father though.
-
olegil wrote:
However, prediction is never accurate, even more so when it comes to predicting the future
How exactly do you predict the present or the past?! :)
-
@mips_proc
What would you do with that stupendous amount of CPU power??
(C&C: Generals is the only SW bit that joe ordinary might need that kind of single-box CPU power for)
-
well what I do for a living is 3D relavent to the print industry/web industry now... so I'd use it for that ;)... I'd love to ditch windows...but its nessisary right now... I'd love to be on OSX it's the best thing apple ever made...but until they get a decent cpu for their machines they leave me relegated to the X86/Wintel world.
-
You didn't read the comment I edited in, did you?
Editing like that is like shooting a moving target… Have you considered “preview” option?
I feel I properly answered BOTH of those questions BEFORE you asked them.
Ok, I’ll do it again with your second version of your post.
What's interesting to note about the 970 is that it's a cut-down high-clock version of the POWER4 series of CPUs, which has already proven itself in battle over a number of years. Both the 64 bit X86 alternatives have so far failed to reach the market (unless you count 900MHz Itanium 2 chips for $2700 as reaching the market).
That would be false in regards to AMD's Opteron. Refer to AMD's "beachhead" programs (some sort of “early bird” program). Note that Athlon XP can be made to run like (i.e. compatibility mode only) Athlon 64 (i.e. via AMD’s SimNow).
Define “hitting the market" (i.e. you seems to apply different “hitting the market" definitions on different products).
IBM's PPC970 64bit CPU is not the only one to attend the CeBIT show.
Refer to http://www.theinquirer.net/?article=8274
Summary
- Several mass-producing X86 motherboard manufactures will be showing their X86-64 motherboards products i.e. these include Tyan, MSI, Gigabyte and Asus.
- MS's Windows X86-64bit version (beta build) edition will be also on show (not a IA-64 edition).
- AMD Athlon 64 powered Laptops.
Note that “IA-64” wasn't defined as "X86-64" (refer to www.x86-64.com).
However, prediction is never accurate, even more so when it comes to predicting the future.
With the X86 market, one can predict the success of this market due to the past 16bit-to-32bit X86 transitions i.e. 8086/80286 (DOS/Windows 3.11) to 80386/80486/Pentium (Windows9x**/Windows NT 4.0/Windows NT 5.x). As this past transition has shown, Microsoft (the X86 king maker) will play a large part for the success of X86-32bit-to-64bit transitions.
Secondly, the large number of third party X86 motherboard manufactures (includes their corresponding customers) that supports AMD’s X86-64 platform cannot be underestimate since they may well dictate the next ruler of the massive X86 Empire. One doesn’t expect 90 percent of desktop users (including the supporting industry) to dump their software investments or move to a graded performance (for legacy software).
Both Transmeta*** and AMD support X86-64 initiatives. ***Insurance as a second source for X86-64 platform.
RAMBUS mess has shown the limits of X86 motherboard manufactures (i.e. their steadfast support for SDRAM and DDR SDRAM technologies) to follow Intel.
If the 970 can indeed hit the market in reasonable clock ranges at a reasonable price it will most probably be something of a success.
Define “success”. It’s not the first time that the PPC has the same clock speed as with X86 CPUs (refer to 1995 era). I’m just too old for such optimistic appraisals.
But how much really depends on how Intel/AMD go from here. Do they continue to produce faster and faster 32 bit chips, or start migrating people over to 64 bit?
You are forgetting one critical factor i.e. the Microsoft factor. From the average Joe and Jane POV (i.e. 90 percent of desktop market); “Microsoft” as a name is a powerful brand name.
Do they continue to produce faster and faster 32 bit chips, or start migrating people over to 64 bit?
Are you implying the Athlon 64/Opterons to be slower than Athlon XP 3000+?
I don’t think there would be Athlon 3600+ based on the current K7 Athlon XP’s Barton core.
Impossible to say. But I really expect Intel to reach something of an upper limit in clock frequencies soon.
Your statement reminds me of mid-90’s “Intel has reached its limit rhetoric”. Some people never learn (sigh).
Edit: I would just like to comment that I'm not saying POWER4 ain't been as expensive as Itanium 1 and 2. I meant to make the point that the 970 is a more marketable version of a battle-proofed chip, whereas Itanium 2 is still only for HIGH END
Such proclamations don’t hold much water in regards to “is still only for HIGH END”. The potential for 100+ transistors silicon chips for mainstream desktop use is not impossible e.g. refer to nVidia’s GeForce FX as an example.
Like DEC’s Alpha, the Itanium has battle X86’s massive empire’s legions of users, support groups and distribution channels. The 64bit distinctiveness would be yet another distinctiveness to be assimilated within X86 collective.
Hmm, the "Unverified source" story?
Insurance.
-
mdma wrote:
Note that IA-64 was based HP's PA-RISC. Secondly, what was the price for Power 4 CPUs again?
Weren't CBM thinking of making next gen Amiga's using PA-RISC processors?
Yes that was the plan for a while, AFAIR.
-
Even if we bloated os4 or mos, 2.5ghz would be a heavy overkill on our beloved oS's...i cant even imagine what we could use such an power for atm...first of all we need an OS...then some games and apps....then we can think of this power .)
cheers
-
X86-64 will probably be marketed right at the desktop from the get-go it's a desktop chip...
Initially, server and high performance workstations.
it isnt here yet...and nor will there be X86-Win2K/XP for it initially when it arrives that takes advantage of it bieng 64bit... aswell as no X86-64 apps initially.
Note that the official release of AMD Opteron and MS Windows Server 2003 (a.k.a dotNET Sever) will be around April 2003 (PS; the planned release, delays can happen).
Note that MS Windows 2003 Server replaces MS Windows 2000 Server range (Win2k code base).
There would be X86-64 OS and applications i.e. SUSE Linux X86-64, Red Hat X86-64 Linux (a.k.a "Red Hat Linux Advanced Server").
To show that AMD64 application and tool’s existence refer to;
http://www.microsoft.com/ddk/debugging/installAMDbeta.asp
MS Windows 2003 AMD64 edition is currently being beta tested at this time.
If we use the past as any indication, the next MS Windows release would be probably plagued by delays.
-
Does AOS 4 support 64 bit extensions, or would AOS on PPC 64 be a total rewrite ?
-
@lempkee
Hmm, lets put it this way:
There is allways a way to make FXPaint or so crawl on
any HW. Done that with the native-x86-plugin and a 1.4GHZ CPU,
and can't say it was hard. Just take a big image and put it
through some heavy filters. :-D
-
filson wrote:
Does AOS 4 support 64 bit extensions, or would AOS on PPC 64 be a total rewrite ?
I recall, Amiga's latest version of FFS was already 64bit (just not 64bit PPC 970 native)...
-
@Hammer
That has nothing to do with OS itself (I'm just assuming
you haven't mad a cheap joke here .....).
64-FFS only means that it can handle bigger than 4GB (32) bit,
moving the OS to accept 64bit wide addresses would break
compability and is simply impossible for the existing apps.
A "boxed" design would be needed or some obscure page-swapping
like it had been done in MS-DOS .....
-
en if we bloated os4 or mos, 2.5ghz would be a heavy overkill on our beloved oS's..
Are you implying we just run the OS without the applications?
.i cant even imagine what we could use such an power for atm...first of all we need an OS...then some games and apps....then we can think of this power .)
Extra speed may open new possibilities for the software vendor’s creative energy i.e. new software release with more elaborate plug-ins.
-
lempkee wrote:
Even if we bloated os4 or mos, 2.5ghz would be a heavy overkill on our beloved oS's...i cant even imagine what we could use such an power for atm...first of all we need an OS...then some games and apps....then we can think of this power .)
cheers
I partly agree (but you should perhaps say "overLIFE" instead ;-)). As a "normal" user you will get very far with just a G3 @ 600MHz when using a lean and unbloated OS. That is one big thing that differs AmigaOS/MorphOS from Windows. On the Wintel platform the situation is different. Intel/AMD releases a new powerful CPU and then Microsoft releases a new even more bloated version of windows that sucks up all the juice it can get. So in the end the user find himself using a new, powerful, system that in real life feels no faster than his old one.
But even on a lean and unbloated OS there will allways be those applications that simply can not get enough of horse power, such as 3D rendering software, so ...
-
64-FFS only means that it can handle bigger than 4GB (32) bit, moving the OS to accept 64bit wide addresses would break compability and is simply impossible for the existing apps.
Of course i.e. I do know the concepts of WOW (Window on Window) compatibility layer.
The sandbox approach is another method to gain compatibility with legacy applications.
PS; Pictures of PPC 970 and its corresponding motherboard would be nice.
-
PPC 970 can run both in 64-bit mode and 32-bit mode, and in 64-bit mode you can run both new 64-bit applications, and old 32-bit applications.
I expect the cost of the 970 to be quite reasonable myself - the processor is a spin-off of the Power4, the design is quite automated and quick, and the die size isn't that large. It shouldn't be much more than a fast G4 in price according to some people on Mac forums, perhaps even less if the yields are high.
So more $400 than $800, in my opinion.
-
PPC 970 can run both in 64-bit mode and 32-bit mode, and in 64-bit mode you can run both new 64-bit applications,
Why not Power 4's applications (i.e. run IBM server software on the cheap (relative) )?
-
@Hattig
good to hear... lets hope its performance is 'decent' at least.
-
Tickly wrote:
olegil wrote:
However, prediction is never accurate, even more so when it comes to predicting the future
How exactly do you predict the present or the past?! :)
You've never tried:
1: archeology
2: history
?
Remember that "to the victor belong the spoils", hence what you see in history books about the past is what someone write in the book, not necessarily what happened. But predicting the future is harder still :-)
-
mips_proc wrote:
well what I do for a living is 3D relavent to the print industry/web industry now... so I'd use it for that...
Ok.
I hope Hyperion manages to get Realsoft3D ready soon after AOS4.0 release... and Realsoft3D comes with renderfarm support. :-)
-
nahh I'll be getting a mac if anything for real world ... realsoft3D dosent float my boat ;P
-
How exactly do you predict the present or the past?! :)
You've never tried:
1: archeology
2: history
You've never tried a dictionary? :-P
[color=006600]pre·dict[/color]
To state, tell about, or make known in advance, especially on the basis of special knowledge.
v. intr.
To foretell something; prophesy.
[color=006600]Synonyms:[/color] predict, call, forecast, foretell, prognosticate
These verbs mean to tell about something in advance of its occurrence by means of special knowledge or inference: predict an eclipse; couldn't call the outcome of the game; forecasting the weather; foretold events that would happen; prognosticating a rebellion.
-
@kronos
A "boxed" design would be needed or some obscure page-swapping
like it had been done in MS-DOS .....
I've read that the 970 is compatible with 32bit apps and that it clears the unused extra 32 bits.
-
@crumb
Which won't be a prob solong you have "only" 4GB of RAM.
But imagine this:
A 32bit-app running in the lower 4GB.
A 64bit-app with data at 0x0000000100000000.
Those 2 want to communicate with each other ....
Mixing different address-sizes is not a good idea, and the
older version of Windows are a perfect proove for it.
-
mips_proc wrote:
PPC970 looks interesting now that the top speed will be 2.5ghz and not 1.8... at 2.5ghz it might hold the speed crown for a time.
:-o
-
Weren't CBM thinking of making next gen Amiga's using PA-RISC processors?
Yes and no. CBM was using the PA-RISC CPU core in their Hombre chipset. However, there is a scenario where CBM could have wound up producing whole CPU's. HP was looking for a company to jointly develop a lower-end version of it's PA's for the consumer/embedded range at this time, and CBM was an ideal matchup. Making CBM an even better canidate over the competition (Intel winning out in the end) was the fact that they were already designing a consumer-oriented version of the PA-RISC Core for the Hombre chipset.(which, frankly, would have rocked and even today would not have been too out-of-place in a budget desktop, as you'll find with Voodoo's and i810 chips today) CBM, if the Hombre design did take off, might have expand the partnership to include whole CPU's. Would have been tres cool.
Woulda coulda shoulda, doesn't matter now. I'd still love to see the Hombre's files tho, just to see what could have been.
But, that's just my personal opinion based on outside observation.
-
Keep in mind PowerPC has been 64bit since day 1. Intel and AMD are still out of their league. All I hear about Itanium and Hammer are delays, lackluster performance, delays, and more delays.
The Mac rumor sites have reported that 2.5Ghz PPC 970 was working using the 130nm process, but the chip consumed as much power as a 3Ghz P4. Of course the PPC 970 does more per cycle...yum! Question is can one supply power to these chips in dual or quad configuration :-o
-
Only a rumour I hope. What is the point of new CPUs using up more
energy - I'd rather keep the old one for desktop use.
-
Quote:
Weren't CBM thinking of making next gen Amiga's using PA-RISC processors?
Yes and no. CBM was using the PA-RISC CPU core in their Hombre chipset. However, there is a scenario where CBM could have wound up producing whole CPU's. HP was looking for a company to jointly develop a lower-end version of it's PA's for the consumer/embedded range at this time, and CBM was an ideal matchup. Making CBM an even better canidate over the competition (Intel winning out in the end) was the fact that they were already designing a consumer-oriented version of the PA-RISC Core for the Hombre chipset.(which, frankly, would have rocked and even today would not have been too out-of-place in a budget desktop, as you'll find with Voodoo's and i810 chips today) CBM, if the Hombre design did take off, might have expand the partnership to include whole CPU's. Would have been tres cool.
Woulda coulda shoulda, doesn't matter now. I'd still love to see the Hombre's files tho, just to see what could have been.
But, that's just my personal opinion based on outside observation.
Sounds interesting, what was the Hombre chipset then? AAA? Anyyone know of any links with info?
-
Sounds interesting, what was the Hombre chipset then? AAA? Anyyone know of any links with info?
it was not AAA, it was post-AAA. They began Hombre when it was obvious that AAA had fallen too far behind. (actually, the origins of Hombre began with an object study to see about putting AAA and a 68k processor into a single chip for a laptop, but things evolved far since that original concept) it did not have much in the way of legacy support for OCS/ECS/AGA. there's a good review of it at http://amiga.emugaming.com/hombre.html (http://amiga.emugaming.com/hombre.html).
There are a few mistakes in that review, however, as some legacy compatability did exist. Most notably the fact that Hombre still used the AAA's audio/peripheral chip, Mary. In addition, it still supported playfields and had added blitter modes to make a dedicated sprite engine redundant. it did keep one sprite tho, Sprite0, which was now slaved as a mouse pointer.
Hombre on a PCI card would have also netted Commodore sales outside of the Amiga platform, as Hombre would have compared well with the top-end chipsets from other vendors and came in at a much lower cost *and* provided features such as texture mapping and dynamic lighting that the other vendors did not add till '97 (when Hombre was slated for production in '95).
but, it's all just a dead past now. A shame, really.
-
Maybe CBM should've sold out to 3dfx or NVidia, things could've been extremley different
-
Define ?success?. It?s not the first time that the PPC has the same clock speed as with X86 CPUs (refer to 1995 era). I?m just too old for such optimistic appraisals.
Well, if you listen to IBM you would wonder if this is the same "x86" Big Blue 20 years ago. IBM has restructured heavily though it is still considered one of the top copmanies in BOTH hardware and software.
And IBM is pushing PPC rather strongly nowdays. They 've said that noone will speak about Itanium in a few years. And, believe it, if there is any company with the power to do it IBM would be a strong canditate. Plus, don't forget the new chip they are developing with Toshiba and Sony for PS3. That means strong sales for their semiconductor division.
PowerPC is a very good technology - much superior than anything AMD or Intel have in ming. It can scale up to servers or down to PDAs. The future is bright.
-
Well, if you listen to IBM you would wonder if this is the same "x86" Big Blue 20 years ago.
I well aware of IBM’s products and their ever trying to re-control of Personal Computer market i.e.
1. PS/2's MCA is an attempt to re-control X86 market. Results: Failed. VL-BUS and PCI eventually killed it.
2. OS/2 Warp, one of the first pure 32bit Desktop OS for X86. Results: Still lost to 16/32bit kit-bashed Windows 95 and market power of Microsoft.
3. PowerPC initiative (during PPC 601 era), IBM has finally concluded that it has totally lost control of X86 market and decided to create itself a new PC market i.e. "PowerPC" (based on it’s POWER series lines).
4. IBM's support for Linux bandwagon i.e. aimed against some certain OS ex-partner.
I did purchased (for business purposes) IBM hardware (and OS products) since IBM PS/2 Model
55/56 era (late 80s) until the recent NetVista range.
IBM PS/2 Model 56 era was bundled with OS/2 (still have my pristine IBM PS/2 Model 56 user manual).
The purchasing of the Amiga 500 and the Amiga 3000 is just a personal hobby of mine.
Well, if you listen to IBM
The “you listen to IBM” is irrelevant in this case. I don’t have any feelings towards any optimistic appraisal based on "what we’re going to do speech".
you would wonder if this is the same "x86" Big Blue 20 years ago.
IBM has been involved with non-X86 CPUs way back in 1990s (e.g. RISC System/6000 family of workstations and servers).
And IBM is pushing PPC rather strongly nowdays.
The “PowerPC” (short for “Power Personal Computer”) was targeted for “personal computing” ever since PowerPC 601.
I still remember PPC 601 vs Pentium Class cubed texture map demonstration test. PPC 601 was shown to be faster compared to the similar clocked Pentium Classic.
"IBM is pushing PPC rather strongly nowdays" would be closer to IBM's attempt to re-energizing PowerPC range.
Personally, I’ll go for “wait an see”, but past deeds plays a large part in terms market acceptance (e.g. desktop PC space).
IBM has restructured heavily though it is still considered one of the top copmanies in BOTH hardware and software.
So? Any company can “restructure” btw…
still considered one of the top copmanies in BOTH hardware and software.
So? Please note that IBM has to compete with similar level competitors.
They 've said that noone will speak about Itanium in a few years.
Until Intel has exhausted it’s X86 revenues base,
Intel will pump massive $$$ into it’s IA-64 project.
Intel is too well cashed up for this type of battle.
And, believe it,
I'm too old for such things. I’m an atheist in such things..
if there is any company with the power to do it IBM would be a strong canditate.
What power? The power like the OS/2 Warp scenario?
Plus, don't forget the new chip they are developing with Toshiba and Sony for PS3.
So? Should one start a list who supports X86 market?
That means strong sales for their semiconductor division.
Good for the group.
PowerPC is a very good technology - much superior than anything AMD or Intel have in ming.
???ming???
Careful with these types of statements i.e. your statement is open to counter attacks.
Did you forget X86 market is consist of multiple companies?
It can scale up to servers or down to PDAs. The future is bright.
Oh boy...
-
strobe wrote:
Keep in mind PowerPC has been 64bit since day 1.
That would be the Power Series not “PowerPC”…
Intel and AMD are still out of their league.
Just like during pre-386/pre-Win32 days (before it bashed the existing 32bit desktop platforms out of the water)...
All I hear about Itanium and Hammer are delays, lackluster performance
Again, where did you get "lackluster performance" for Itanium and Hammer?
The Mac rumor sites have reported that 2.5Ghz PPC 970 was working using the 130nm process, but the chip consumed as much power as a 3Ghz P4. Of course the PPC 970 does more per cycle...yum! Question is can one supply power to these chips in dual or quad configuration :-o
What was the release date for this product?
-
I think X86-64 (Opteron) is definitly the overall future, that chip is going to be incredible aswell, it's got an integrated memory controller wich makes SMP bandwidth multiply instead of devide.
-
The PowerPC has an instruction subset of POWER. POWER is basically PowerPC+Amazon.
PowerPC was 64bit from day 1. The PowerPC 640 may have been the first 64bit chip with only PowerPC instructions, but 64bit POWER processors existed beforehand which have the same instruction set.
Basically what I'm saying is you're wrong.
As for 32bit performance, x86 only caught up to the 68000 series when they released the 486DX4, and it didn't exactly blast anything out of any water (although it could boil water).
Lackluster performance refers to IA-64. Hammer really isn't 'here' yet, it's in one of its many delay cycles.
The release date of the PPC 970 will likely be late summer early fall. Plenty of time for more delays from the 64bit wannabes.
-
by mdma on 2003/3/13 6:49:51
Weren't CBM thinking of making next gen Amiga's using PA-RISC processors?
Crap, I see Downix already beat me to this one. Oh well.
Dammy
-
The PowerPC 640 may have been the first 64bit chip with only PowerPC instructions,
May have been?
A few of the PowerPC CPUs wasn't even pure 64bit CPUs i.e.
PowerPC 602 Embedded Processor,
PowerPC 603e Microprocessor,
PowerPC EM603e Microprocessor
PowerPC 604e Microprocessor,
PowerPC 740 Microprocessor (300MHz to 550MHz),
PowerPC 740 Microprocessor (up to 266MHz),
PowerPC 750 Microprocessor (300MHz to 550MHz),
PowerPC 750 Microprocessor (up to 266MHz),
PowerPC 750CX and 750CXe Microprocessor,
PowerPC 750FX Microprocessor,
-
Hmmmm...from what I can find, the POWER
(Performance Optimization with Enhanced
Risc) architecture was not fully 64 bit, but
its FPU did indeed have 32 64 bit registers
(branch and integer registers remained 32 bit).
I'm not sure whether or not the 64 bit FPU
architecture was carried over to the PPC
line of processors before the 620 (the first
true 64 bit PPC) but if so, it may explain why
the early 66mhz 601's stomped the 66mhz
'060's in the FPU benchmarks, while remaining
relatively on - par or even slightly slower
in other benchmark areas.
-
As for 32bit performance, x86 only caught up to the 68000 series when they released the 486DX4,
So?
Why did DID cut down TFX for the AmigaOS platform release?
"X86 only caught up to the 68000 series when they released the 486DX4" generalization would be false when given a certain scenario.
Are you saying a 68030@25Mhz can beat 80486DX @33Mhz?
Are you referring to 68060@66Mhz?
Please be more specific.
Lackluster performance refers to IA-64
Are you referring to IA-64's 32bit performance?
Hammer really isn't 'here' yet, it's in one of its many delay cycles.
What about the delays? Most of release dates are just estimations.
At the moment; AMD was still in game (against Intel) with their current Athlon XP’s Barton core.
-
give IA64 a few years... it's comming slow and it's comming big... Intel is nobody fool...they have enough money to license PowerPC if they needed it... their filthy rich... and they wont be beaten so easily...
People often concentrate on how much X86 sucks (wich is dosent, benchmarks show otherwise) while never concentrating on it's price/performance ratio...and the fact that it's backward compatible to nearly 30 years of code (first x86 is 1976?) ...
Intel's done a damn fine job of SUPPORTING it's customer base for such a long period of time... meanwhile motorola just outright dumpes 68K and dosent give anyone any other option except to dump it... no backward compatability... no forward momentum...
now here we sit 10 years or so later... and we got 'UNDERPERFORMING" (compared to X86) PowerPC on the 'desktop'...
so I ask... what benefit did PowerPC give them?... If intel had totally dumped X86 and gotten to start from scratch... would they have a CPU that's architecture is damn near a junker 10 years later? I doubt it... thats bad planning... thats bad design... good design allows for future developments... not middle-of-the-night bail outs from big blue...
Intel's done a damn fine job creating cheap, relatively reliable processors that are damn near 30 years backward compatible (wich MATTERS to business).
I wont knock them or put them out of the game so fast. IA64 might suck right now...but in 5 years... I'll bet right now... we'll all be sitting here arguing over benchmarks where the IA64 is competing with the X86-64 and both Opteron and Itanium are stomping the PunyPC2200 or whatever they call it.
I'm going to buy a Mac probably as my next machine...and I dont knock big blue's engineering it's got alot more faith from me then Motorola's does... but I'm realistic... when Intel starts going in a direction... their such a driving force that (the entire market really) there is nothing stopping them... except time ... wich holds them back... IA64 or whatever they choose for the 64bit market will be big... it will take time...and AMD/IBM can make hay while the sun shines... but it wont shine forever.... AMD has a good chance at taking a big percentage of the market and comming on par with Intel... but Intel's got full bank accounts and a very broad spectrum and such a high percentage of the overall market that it's going to hard to derail them... Intel would probably sooner make thier own X86-64 then give that market away to AMD...
-
Until Intel has exhausted it?s X86 revenues base,
Intel will pump massive $$$ into it?s IA-64 project.
Intel is too well cashed up for this type of battle.
Rephrase: "They 've been a failure though they will be the last to admit it".
So? Please note that IBM has to compete with similar level competitors.
Few can be credited being right at the top in BOTH hardware and software.
So? Should one start a list who supports X86 market?
Let me guess. Dell ?
Well, just about everybody uses x86 nowdays, including Sun, Sony, Toshiba and ... (aargh !!) IBM. But these companies (unlike Dell and Gateway) are not that tightly dependent to Intel for their revenues.
What power? The power like the OS/2 Warp scenario?
Well at least the try. Intel is good in two things:
1) build and supply many CPUs
2) overhype their products
They are going nowhere with Itanium because there is no market for it in the desktop (i.e. no 64bit Windoze) and they have to rely in others (HP) in the server market where just selling hardware is not enough.
-
Well at least the try. Intel is good in two things:
I'd put 'supporting their architecture' in there aswell... wich is one of the BIGGEST factors to business... they ask 'will there be OS's and software for this in 10 years?" with X86 the answer is still yes.
-
Just a few facts :
1) Opteron will reach the market by april 22, when AMD announce them, provided it is not a paper launch and you can effectively buy them (at premium price in server boxes).
2) Athlon 64 (the consumer version) will reach the market in septemeber, if all things go well, by that time, it would be possible to see ppc 970 chips chipping in apple products, but this may be a bit optimist.
3) ia64 is a joke, intel sold 2500 servers running ia64 in 2002, worldwide.
4) IBM helped AMD out with their copper + SOI process, and IBM is said to have the fastest from design to market time, so it is well possible to have ppc 970 before athlon 64 (which is already 6 or so month late).
5) itanium 1, which didn't exceed 800 MHz or so was a total failure, and not designed by the HP team. itanium 2, was designed by the itanium team.
6) about price, you cannot compare the price of the high-end dual core power4 with that of the consumer-level ppc970. Most power4 come in 4 CPU/8 cores CPU daughterboards with insane amount of L3 cache, there is no way this can be cheap.
Friendly,
Sven Luther
-
Rephrase: "They 've been a failure though they will be the last to admit it".
For Intel’s POV it would be just a "black eye". It’s not the first time that Intel has embarked on non-X86 processors (not including StrongARM RISC families) venture.
Few can be credited being right at the top in BOTH hardware and software.
Did that combination change their fortunes in regards to the desktop PC market?
Let me guess. Dell ?
Did you forget the following supporters?
Micro-Star-International, Tyan, ASUStek, Gigabtye, nVidia(e.g. nForce2 x86 chipsets; a close AMD partner), ATI (e.g. A3/3x0 x86 chipsets), Aopen, Acer labs (Mitsubishi), Gateway, Walmart, Asrock, Luckystar, Abit, VIA, Chaintech, Evergreen, Supermicro, Transmeta, SIS, Soltek, Newsys, ALi, Microsoft(1), and 'etc'. Just about 90 percent of desktop market vendors and mainstream distribution channels.
Notes;
1. Microsoft has repeatedly stated that it strongly prefers AMD's 64-bit architecture to Intel's IA-64.
2. Red Hat, Mandrake and SUSE has support for AMD64.
For Mandrake's AMD64 Linux refer to
http://www.mandrakesoft.com/products/90/
Intel is good in two things:
1) build and supply many CPUs
2) overhype their products
You are forgetting the other half of the partnership i.e. the Microsoft factor.
@strobe
Refer to "http://www.amdboard.com/hn03130301.html" Athlon 64's K7 rating (Via Canadian Québec distributor MRS MAINTENANCE).
1800Mhz = 3200+ (Planned for June 2003).
1900Mhz = 3400+ (Planned for June 2003).
2000Mhz = 3600+ (Planned for June 2003).
2100Mhz = 3800+ (Planned for September 2003).
One could guess the Athlon XP's Barton Core (3200+) is nearing the end of it's life cycle i.e. 400Mhz DDR is the last stop for EV6 architecture**.
Such a scenario would be similar to 486 days i.e. running 16bit OS (e.g. MS-DOS/MS Win3.11) on 32bit hardware (Not factoring MS Windows 2003 Server AMD64 edition, Red Hat Linux Advance Server AMD64 Edition and ‘etc’ releases).
-
Sven highlighted a few important points up there, which I must say ties in pretty well with my initial statement somewhere on the first page of this thread. PPC970 will be a consumer-level CPU based on the POWER4 architecture, and if IBM get it out on the market in time it's got all the chances in the world of becoming a success. Because neither Intel nor AMD have a stronghold here yet.
-
Sven highlighted a few important points up there, which I must say ties in pretty well with my initial statement somewhere on the first page of this thread. PPC970 will be a consumer-level CPU based on the POWER4 architecture, and if IBM get it out on the market in time it's got all the chances in the world of becoming a success. Because neither Intel nor AMD have a stronghold here yet.
Define success in your view.
Bringing out a CPU product alone doesn’t automatically equal success. Many other factors will aid the level of success for the particular product.
Did you forget 64bit MIPS based processor?
Because neither Intel nor AMD have a stronghold here yet.
It didn't stop Intel from reaching world domination during the late introduction of it’s first X86-32bit processor (e.g. 80386DX).
Intel alone (and X86 market as a whole) did not reach world domination by it’s own abilities i.e. one must the Microsoft factor.
-
I agree with mips_proc...Intel after all
invented the first single - chip cpu...and AMD
goes back to the '70's as well. Intel has
supported the x86 architecture for 20 years,
even with built - in chip emulation of the
older instructions.
They're not going anywhere like Motorola,
who left several companies and user
bases "in the water" when they 'bailed' on
68k and other architectures.