It's (Almost) Hammer Time 344
thelizman writes "C|Net is catching up on the buzz with AMD's Hammer line of processors. Of note in the article is how AMD demonstrated their 64-bit contender using Linux and Windows, instead of just Windows. In reality, Linux will likely have 64 bit applications more quickly than Microsoft, and will see use on this processor more readily than your average WinTel machine, so you know...like...it only makes sense."
Showcasing Linux shows... (Score:2, Funny)
Re:AMD, i love you. (Score:2)
yeah, right. Just like the K-6
sure, our processor supports it. No, noone ever made a chip set, though . .
hawk
Re:AMD, i love you. (Score:2, Informative)
Re:AMD, i love you. (Score:2)
I believe the hammer will ship. I'll believe in SMP hammer when a vendor demonstrates one . .
besides, it was a tongue-in-cheek wisecrack.
hawk
Re:Threat to Sun (Score:2)
Re:Threat to Sun (Score:2)
1) The numbers you quote are "peak"numbers, obtained by optimizing for the particular test being run. The "baseline" numbers are probably a better approximation to what you'll see in real world use. Here, the story is reversed: The Dell scores 779 vs. the Sun's 701.
2) SPECFP? What about integer performance, which is more important for most applications. On SPECINT2000, the Dell whips the SunBlade both in baseline (790 vs. 537) and peak (811 vs. 610) performance measurements.
3) How about price/performance? I can't find any mention of the Sun Blade 2050 on Sun's site. However, the 900 MHz Sun Blade 1000 (which is slower than the 2050) goes for $11,000. I can get the Dell with similar amounts of memory and HD space for $2500.
4) How does the Sparc's better performance / clock make it a better CPU? Is there some intrinsic value to clock cycles that I don't know about?
Look, I think there are good reasons in some cases to buy Suns over commodity Intel hardware. And there's probably a good argument to be made that these benchmarks don't correspond well to any real world performance, anyways. But you're just fooling yourself if you think that Sun beats out Intel when it comes to raw performance as measured by benchmarks.
Re:WHAT!?!? IDIOT MODERATORS (Score:2)
And yes, I thought your pun was funny.
news for nerds... (Score:1, Funny)
Will this hammer... (Score:4, Funny)
Re:Will this hammer... (Score:5, Funny)
With no cooler:
Can't thouch this!!
Re:Will this hammer... (Score:5, Informative)
cf: IA64 (Score:5, Interesting)
In general, I doubt strongly this is a AMD vs Intel issue, either. This is a Windows (and their legacy users) vs Linux (and their overly prideful users that must find every method to berate windows).
Re:cf: IA64 (Score:5, Informative)
The site also has a 64-bit simulator [x86-64.org] for you favorite 32-bit processor based Linux system.
Re:cf: IA64 (Score:2, Insightful)
Don't forget the folks who claim that a particular stereotype perpetrated by a few zealots must obviously apply to an entire class of people....
Max
AMD's New Slogan (Score:3, Funny)
Talisman
Re:AMD's New Slogan (Score:1)
touching an AMD (Score:1)
Windows at disadvantage? (Score:5, Interesting)
There are probably enough people like me that don't want to upgrade to WinXP just for 64-bit (I don't like lots of things about XP, but thats my opinion). So it would seem that Linux with Cross-platform portability (hence, x86-64) will have a better chance at propagating (spelling?) itself in to this market faster than windows.
Just my opinions, not to be taken as fact.
Re:Windows at disadvantage? (Score:3, Informative)
AFAIK:
- NT code isn't 64 bit safe. 2000/XP I'm not sure of.
- the 64 bit port of NT was developed on the Alpha, initially anyway, and then ported to the Itanium.
- Alpha Linux has always been 64 bit. One of the earlier kernels had to be extensively revised to be 64 bit safe in order to run on the Alpha.
Soko
(O/T - The Alphas still killed the Intel machines at the time with MHz as well as memory and I/O bandwidth, which is why we used them. Oh well.)
Re:Windows at disadvantage? (Score:2)
Re:Windows at disadvantage? (Score:5, Insightful)
In the transition to 64 bits, if AMD can get there faster (and by there, I mean readily available to the consumer, not readily available to the bored millionaire), they can enlist Linux as their Microsoft and do the same thing to the market that has been happening for a decade: only with a free OS.
Actually, I wouldn't mind, and I don't think many would.
Re:Windows at disadvantage? (Score:2)
It's not nearly so subtle as that, it's recognition of Linux's huge position in the server market, where prices are high and 64 bits is a significant win for file caching.
Re:Windows at disadvantage? (Score:5, Insightful)
Key applications for 64bit computing are more or less involved with anything that requires a huge amount of memory. Servers(massive databases), high-end engineering(airplanes, ships, etc.) and scientific computing come into my mind.
In these kind of applications and systems you're not concerned whether or not you like windows xp but rather: how cost effective is it and what is the performance advantage?
Unless your computers memory capacity is exhausted(what, 4 gigs is not enough for everyone?) and it is crunching numbers on full load 24/7 I don't see too many reasons aside the coolness factor to even consider 64bit computing. Heck, smp systems would make much more sense in most of the cases.
After all, Bill Gates Himself said... (Score:3, Funny)
-Erik
Re:Windows at disadvantage? (Score:2)
I'm not saying that any of these applications couldn't benefit from 64bit processing and the extra registers that come with it but rather: it is going to be a lot more reasonable(cost effective) for at least few more years to come to stretch the limits of 32bit platforms in applications where that 4gig limit is not unmanageable.
Actually, a lot of these applications are primarily going to benefit from increased memory bandwidth. Once the memory bandwidth equals cpu's fsb we could see full cpu i/o utilisation. This does not really happen even with 32bit platforms yet(Never mind when doubling the bandwidth requirements).
Re:Windows at disadvantage? (Score:3, Informative)
It's been pointed out for ages in the NT Insider Newsletter.
My guess is: Microsoft doesn't work in a fishbowl like the Mozilla team does; but it must not cost them much to keep an IA-64->x86-64 port of XP64 ready, just in case (especially since I guesstimate the HAL should merely be a hybrid of x86 and IA-64, the compiler an extension of the x86 logic (much less difficult than VLIW and much well understood), and the code above HAL, once 64-bit clean, is (reportedly) written in compiled, not assembled, languages).
Re:FUD Sucks (Score:3, Interesting)
Re:Windows at disadvantage? (Score:2)
Windows: I can pay for a Windows license for what the OS/2-alone license costs! And the Windows programs run faster, since I'm not running OS/2 in the background!
--
AMD: I can pay for a top-of-the-line x86 processor that also does 64 bits!
Sun: I can pay for a top-of-the-line 64-bit processor!
Intel: I can pay for a rather mediocre 64-bit processor that runs x86 code slower than my current computer, or I can pay for a x86 processor that only does 32 bits.
--
hammer time (Score:2, Insightful)
Re:hammer time (Score:2, Informative)
I plan to get a 2 processor Clawhammer box myself, it's the only reason I haven't upgraded for the past year. I'm bored of having a mainstream PC (P3 550MHz, don't ask...) after using a StrongARM/NetBSD box for a few years. Time for something novel and exciting - dual processor new fangled chip sounds like just the thing...
Cuz its AMD... (Score:2, Interesting)
Not to mention, 64 bit processing on a desktop would be reason enough for me to quit putting it off!
Jason
Memory (Score:2)
It also helps for file size issues as well as addressable partition size issues, all can now be done natively rather then as hacks.
Do you download DVDs or Encode? (Score:2)
When you start doing file sharing by the gigs, and yes you will with a fast machine and you want to handle those files, you'll have problems when your machine can only handle certain file sizes.
More bits also means programs can do stuff like encode big files faster.
Why delay the hybrid? (Score:3, Interesting)
I do like the fact that AMD is planning on using "a smooth migration path to the 64-bit software of tomorrow", so we wont have to rewrite much of anything. Besides, I still like my old DOS games
Re:Why delay the hybrid? (Score:1)
Re:Why delay the hybrid? (Score:2)
That's where linux has the advantage. Todays progs can jump straight to 64bit by the standard
make
make install
So maybe it isn't point and click, but it is 64bit clean and ready to run.
Re:Why delay the hybrid? (Score:2)
According to this, long is going to be 64 bits.
I don't like that, actually. You can't really avoid problems with programs that make assumptions about data sizes, since you are basically stuck with 64 bit pointers, and thus some data size relationship has to change. But for many that use long to define a 32-bit type, they are suddenly going to get 64-bit types when they don't need it. That has the potential to do things like blow out your cache, which would hurt performance.
For well-behaved programs, there shouldn't be any problem recompiling for 64-bit mode. There are some advantages to doing so, like being able to take advantage of the extra registers. For poorly-behaved programs... Well, just leave them in 32-bit mode, and they'll run just as well as before.
Re:Why delay the hybrid? (Score:2, Interesting)
Honestly, IMHO it seems that hybrid or "bridge" products meant to serve as vaseline for new technologies (allowing you to ease into it : ) usually wind up delaying the newer technologies and adding cost to the eventual transition. Apple did'nt write OS-X up to handle MacOS 9.1 apps, and in less than a year they've caught up with core products to an enthusiastic response from Mac users everywhere.
The Hybrid isn't delayed (Score:3, Informative)
Re:The Hybrid isn't delayed (Score:3, Informative)
*cough* nitpick: x86 began as a 16-bit architecture; when you say "8-bit" you're probably thinking of the 8088 which had an 8-bit external data bus and which IBM used in their PC because, basically, they were doing their motherboards on the cheap.
64-bit on the desktop? (Score:3, Interesting)
I don't do much 3D rendering other than some gaming action, and my multimedia is limited to playing some MP3s while I'm coding with vim. Are there any other compelling reasons for a 64-bit arch? I suppose I could load more data in registers, storing two 32-bits into one 64-bit register.... but i'm drawing a blank... someone help :)
Re:64-bit on the desktop? (Score:3, Funny)
So is 64-bit for a normal user going to do much?
Extra computer power will always find a way to get used up in frivolous ways by the sex trade, trust me.
Re:64-bit on the desktop? (Score:2)
Interesting!
I never knew Microsoft was in the sex trade.
Re:64-bit on the desktop? (Score:2)
When you deal with Microsoft, you just know that you're going to get screwed.
Re:64-bit on the desktop? (Score:3, Funny)
Of course -- you can double your RC5 throughput and dazzle your friends. :)
Normal users dont need ghz cpus (Score:3, Interesting)
Well the normal could do fine with a pentium 200 and 64 megs of ram.
64bit is for the power user, people who want gigs of ram, huge harddrives, people who trade media like dvd movies, who edit movies, who play games, who run alot of programs at the same time, or who just want more speed, they want state of the art.
Re:64-bit on the desktop? (Score:3, Informative)
- it's fast
- there are additional registers available which should help compilers quite a lot (avoiding false dependencies: more opportunities for executing more instructions at the same time)
- it's fast.
Ok maybe you could say that you don't need such speed, but the games you play don't look like Final Fantasy (the movie) and your opponents could really be smarter and I suspect that a good AI is very,very CPU-consuming.
It will be "free" (Score:2)
Re:64-bit on the desktop? (Score:2, Insightful)
Re:64-bit on the desktop? (Score:2, Insightful)
I really have to disagree with you there. Computer power and graphics power are so far out ahead of what programmers are writing, it's rather sad.
Just look around at graphical interfaces on computers. X11, Windows, etc. None of these run nearly as well as they should. {clicks to open a new netscape window and waits while the hard drive grinds away, geez 2 CPUs, ultra2 scsi, dual TNT2 cards, 1gig of ram and here I sit *grind grind*}.
I agree that we will always need more power. More power to crunch through the, even more, bloated software of tomorrow. Please don't assume I am being flamebait here -- Just look at how little has changed from the first versions of MS Office to the modern-day MS Office. Not a whole lot of gain for a whole lot of bloat. This seems to happen across just about every part of the software industry.
Call me on this if I am wrong.. Thanks.
Victor
Are you sure? (Score:2)
Blame the programmer or blame the motherboard?
The speed of your ram is the problem, the harddrive wont grind iff you have gigs of ram and you are using a scsi raid system unless your ram is just slow.
so if your system is fast, why is netscape slow ? Fact is its not a good example of a slow program, nautilus is a good example of a slow program.
Little has changed because theres a monopoly,
Things wont change unless you make changes, join the open source movement and develop something new.
I'll tell you how i'd use the CPU, automation, AI, and stuff like that to make my computer do self healing,to make it solve problems, to the point where i can tell it to find information on say, star trek episode 10 and it automatically opens netscape in the backround runs a few search algorithms and looks for information for my research.
This could be done using an agent.
What programmers want has nothing to do with it (Score:2)
Programmers dont want bloated code, users of the program want big powerful apps. Thats why we have photoshops and netscapes and the like.
With media based apps and media editing, and file sharing its just going to keep raising the bar.
As far as ram, ram speed and harddrive speed are two off the main bottlenecks of a PC, raise the speed of ram, and the speed of the harddrive, have about 16 megs of L2 cache, you'll have a fast computer if the ram is feeding data to the cpu at about 6gigs per second, the CPU displays to the screen instantly, everything would be instantanious, bloated code or not, programs the size of windows will load instantly hell your machine will boot instantly, add a 64bit cpu and you'll be able to edit huge files, add scsi raid and you'll have the transfer rates needed to handle it, and for file sharing on your fibre optic line downloadinng hundreds of gigs a week, your 10 terrabyte drive wont last very long.
Sure this PC may be a few years away, but this PC will be the average PC of a power user within 3 years.
Late (Score:2, Insightful)
AMD's Hammer is the same way. We all wait with bated breath for the new processor to drop, but no one's seen it yet. It's surely not vapor because we know it's on its way, but how long do we need to wait? How far into the future should these things be announced.
Hammer has been announced far too long in the past to be of any interest these days.
Let's wait until it actually gets released and then discuss further.
Re:Late (Score:2)
I hope that with your "don't give it a thought until it's released" attitude you never get promoted to be a manager of some kind. You would suck!
But the real question is... (Score:2, Insightful)
...when will there be motherboards that support it?
---Windows 2000/XP stable? safe? secure? 5 lines of simple C code say otherwise! [zappadoodle.com]
Re:But the real question is... (Score:2)
Re:But the real question is... (Score:2)
I'm sorry you don't like my sig, but if you think being able to crash three major operating systems from an unprivileged account by using printf is "stupid" or equivalent to being able to take down Linux from a rooted box, you must not have much security clue. Anyone can crash any box in about 1 line from a privileged account, there's no fun in that. *nix hasn't had a security hole this bad, AFAIK, since the early 80's (I could be wrong on that though). So I think I'll keep it in my sig till I'm good and ready to take it out.
Thanks for the advice though, free friendly advice is always appreciated. :-)
---Windows 2000/XP stable? safe? secure? 5 lines of simple C code say otherwise! [zappadoodle.com]
Can't touch this (Score:4, Funny)
mov ah, #1
mov ax, #1
mov eax, #1
mov eeax, #1
Seriously, I wonder how they have modified the register addressing field of the instructions to handle this.
Re:Can't touch this (Score:2, Informative)
mov ah, #1
mov ax, #1
mov eax, #1
mov rax, #1
Feel free to read the specs [amd.com].
Re:Can't touch this (Score:2)
mov ax, #1
mov eax, #1
mov eeax, #1
Damn, you know you're around smart people when something like this gets moderated '+4 Funny'.
Cut to the chase (Score:1, Funny)
Offtopic Request to CmdrTaco (Score:3, Insightful)
I'm not sure which is better journalism though... on one end, you're looking more professional by not having stupid 14-year-old-girl talk on the front page. On the other end, you're cutting up someone's quote!
I'd rather have it look nicer.
Re:Offtopic Request to CmdrTaco (Score:2)
Usually it's Taco's "lame garbage" at the end.
Re:Offtopic Request to CmdrTaco (Score:2)
Designing the X86-64 architecture... (Score:5, Informative)
Re:Designing the X86-64 architecture... (Score:1)
32-bit emulation? (Score:1)
Re:32-bit emulation? (Score:1)
There's a Pentium-compatible chip built into every Itanium. I guess that's one approach to "emulation." :-)
---Windows 2000/XP stable? safe? secure? 5 lines of simple C code say otherwise! [zappadoodle.com]
Re:32-bit emulation? (Score:2, Interesting)
Directly from the article:
The Hammer family of processors will differ from other AMD chips--and other Intel processors--in that they will be able to run conventional 32-bit applications found on Windows PCs today as well as 64-bit applications.Perhaps we should read the article before we all run off and post ;-)
AMD's diminishing market advantage (Score:3, Interesting)
Answer: .18 vs .13 techonology (Score:2)
Kjella
AMD was right to grab every DEC Alpha engineer (Score:3, Insightful)
Intel: Buys its way out of a lawsuit for stealing 64bit microcode from the DEC Alpha, then buy's the Alpha from Compaq to discontinue it. Then create a brand new 64 bit chip using their own limited talent, while shoving the existing 64 bit platfrom into an early grave.
Does this make sence to anyone? Alpha's rock, and they have been 64 bit for years. There already was versions of Win2k, Linux and Unix in addition to major apps like SAP and Oracle tuned for the platform.
Re:AMD was right to grab every DEC Alpha engineer (Score:3, Informative)
Re:AMD was right to grab every DEC Alpha engineer (Score:3, Interesting)
AFAIK Alphas died because of business problems, not technical ones.
Re:AMD was right to grab every DEC Alpha engineer (Score:2)
Re:AMD was right to grab every DEC Alpha engineer (Score:2)
Re:AMD was right to grab every DEC Alpha engineer (Score:2)
Re:AMD was right to grab every DEC Alpha engineer (Score:2, Informative)
http://news.com.com/2100-1001-227510.html?legac
There was ONE or two 1GHz Alpha's mounted in a SlotB format at API in *1999*. These had to use 250MHz cache (at a 1:4 ratio). Limitations in the Tsunami chipset didn't favor anything more than an 833MHz. This meant that with the slow cache and Tsunami limitations, the 1GHz was like putting a Corvette engine in a Cavalier. Goes fast doing only one thing. Don't take a corner.
Those one or two 1GHz Slot B's were proto's. They are probably still on my former desk or in the lab or maybe have been shipped back to Korea by now. API closes its doors this week from what I've heard.
I worked at API.
AMD/Intel in the press (Score:2, Insightful)
The AMD article is a simple response to a press release. The Intel article is a prose editorial about the state of the industry and where Intel's new processors (might) fit in.
Re:AMD/Intel in the press (Score:2)
But was it running *64 bit* Windows? (Score:3, Interesting)
The PR is vague enough to be interpreted as "running a 64-bit version of Linux as well as [plain old 32 bit] Microsoft Windows". I've asked AMD flat out, and they will not commit to saying yes, Win64 will be coming to the Hammer party. MS certainly haven't mentioned it, AFAIK.
As a film/video FX developer, we'd love the massive memory space & 64 bit registers that Hammer brings. But as a [currently] Windows-only app, Linux-64 isn't helpful (except possibly for a few customers' render farms).
Our code is 64-bit clean, we have a working Itanium port, but we haven't sold a copy yet. We have customers who need multigigabytes of RAM & the speed of an Athlon to process it all, yet don't have the spare kilobux to justify dedicating a dual Itanium to a single app (it's all but useless for 32 bit apps at Winzip level & up).
So... rumours, anyone? Hard facts? Tidbits, gossip, insider info?
Re:But was it running *64 bit* Windows? (Score:2)
Re:But was it running *64 bit* Windows? (Score:2)
what would qualify you to know wether or not microsoft has windows running on x86-64 ?
do you work at amd ? do you work at MS ? i suspect that at either of those places, you wouldn't be allowed to talk about it if they _did_ have it working.
he asked for insider info on the status of win64 on amdx86-64, and you give him a post about porting his software to linux.
so, what reason do you have to belive that you know what the status/existance of that project may or may not be ?
Re:But was it running *64 bit* Windows? (Score:2)
Re:But was it running *64 bit* Windows? (Score:2)
Thanks anyway :-)
Re:But was it running *64 bit* Windows? (Score:2)
Of course, if a monopoly verdict doesn't encourage MS's readiness to announce support of non-Intel platforms, I don't know what will...
Re:x86-64 support probable (Score:2)
Pathetic (Score:2)
Honest Question. Isn't the bottleneck still Disks (Score:2, Insightful)
As we go over 2GHz, and from 32 to 64 bit, bus speed is going up (good), memory seems to be creeping up on speed (RAM that is)....
But what about hard drive access speeds? They don't seem to be getting faster at the same rate as everything else. And, the only think I seem to ever be "waiting" for using my 32bit 1Ghz system is reading something from the hard drive.
Anandtech article with pictures! (Score:4, Informative)
Anandtech has posted an article with lots of information and pictures Right here [anandtech.com].
Re:Is x86 really the way to go? (Score:3, Insightful)
Uh-Oh (Score:2, Funny)
Re:What's so great about 64 bit? (Score:2)
Three years ago, I thought I was partially insane for getting a laptop with 128 MB of RAM. Turns out my insanity was a good thing.
Today, I'm thinking I'm partially insane for getting a machine with 1 GB of RAM. I'll undoubtedly be congratulating myself for my foresight in a couple years when Windows ZZ requires that much to operate.
Re:What's so great about 64 bit? (Score:2)
Desktop applications requiring lots of RAM (Score:5, Insightful)
1. Non-linear video & film editing:
Current video editing software can work from and to disk, but availability of more RAM will make it easier to do more sophisticated effects in real time.
2. Genome sequence analysis
Okay, not very many people will be doing this, but it IS a growing field, and people are doing the work on desktiop machines now (albeit slowly).
3. Modelling / CAD
You can never have too much memory in a CAD workstation.
4. Software development
Again, you can never have too much memory. More memory enables more agressive optimization, as well as supporting more productivity features in the IDE (like full source indexing). I have used toolsets that need 2+ GB of RAM to compile a relatively simple program (they swap now, of course).
So, probably not for Microsoft Word '03, but there are definitely applications for 64-bit computing out there other than servers.
-Mark
Re:Desktop applications requiring lots of RAM (Score:2, Interesting)
More RAM will not matter if you can not access the data which you desire from the permanent storage device. ATA and SCSI just can not deliver, mostly due to individual disk drive IO. Fiber Channel is close, but not practical, and is no different than something like RAID1 when it comes to performance. It is all stored on disks. That is going to have to change or something. Disks are a mechanical device and are not going to scale. Something is going to have to give. Something new needs to be made to accelerate this IO from the permanent storage into the temporary manipulation space (RAM).
Right now, I do not want a faster processor, because that will not improve the speed of disk IO. I do not need more RAM, because I really do not use it (today). Faster network speed? That would be nice, but the 11MB across my LAN is okay for now.
I just wish that moving that data around was faster, not necessarily being able to hold more of it. And in this case of speed, the processor has nothing to do with it.
Now what this all has to do with this new AMD processor, I do not know. Nothing. Mod me down.
Listen to what you're saying... (Score:2)
It is all stored on disks. That is going to have to change or something. Disks are a mechanical device and are not going to scale.
And if you had 10GB of RAM, how often would you have to read or write something to disk? Almost never. Your "permanent storage" just becomes a back-up in case power fails, and the whole thing just runs from RAM. Disks are never going to be nearly as fast as RAM (they're each optimized in different directions). The best way to improve I/O performance in a computer system is simply to not do any I/O
-Mark
Re:64 bit proc = extreme heat? (Score:2, Informative)
Re:64 bit proc = extreme heat? (Score:2)
Btw, the XP and MP line implements a thermal diode. Your mobo can throttle or shutdown the same way the P4 does if you want, but if you are at all intelligent on your case design, etc. you will never have to do that. Many of the bioses today implement a shutdown temperature driven off the termistor (ECS for one makes mobos that do this).
I have never seen a fan melt off a heatsink. Can't quite imagine how he managed that one.
Re:64 bit proc = extreme heat? (Score:2, Insightful)