With Opteron now officially set to debut at 1.6 and 1.8 GHz clockspeeds (and no 2 GHz model as initially hoped) discussions have resurfaced as to how well AMD is able to scale the Hammer architecture. 1.8 GHz, after all, is nothing new for the AthlonXP? AMD reached this speed nine months ago with the nuclear-furnace original-model 2200+ and shot nimbly past that speed once AMD revised their .13 micron process and cut their heat dissipation.” Read the article at The Inquirer. Athlon64 benchmarks here.
Is AMD Being Coy with its Clockspeeds?
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Wasn’t there supposed to be a thrid one, the 244 to be release pretty soon (may?)? That would be the 2ghz one then.
Also the “benchmarks” on xbitlabs are uhm… pretty useless. They are benchmarks of a pretty old engineering sample (B0), and it’s still about 5 months before the chip (and chipsets) are actualy released. We’ll get some idea of the x86-64 performace on the 22nd (from decent sources, in decent english).
Clockspeed does only give you half the story if even that. Comparing an Opteron with an AthlonXP by their relative clock frequencies is just not interesting. If it worked this way, we would all drive 18-wheelers since they surely must be faster than a fourwheeled car ?
Isn’t it?
As dege already said, clockspeed isn’t the One And Only thing, especially in this case, because Opteron is to be an _Server_ cpu, and what’s more critical in a server, speed or stability?
Those benchmarks seem to be dated the 4/18/2003. Just 2 days ago. Also, they seem to have the A64 1.6Ghz. I think you were referring to the old article, where they benchmarked some engineering sample running at 1 Ghz.
The Athlon64 is the Clawhammer, which is meant to be used in desktops. The Opteron is the one that’s meant for servers.
Well, few things.
1. I just don’t believe in the performance of the AGP or IDE of any VIA chipset, esp. this one, some kind of beta chipset.
2. The Opterons, if you mind, are server/workstation CPUs, the things like many HyperTransport channels and multi-cpu glueless (no additional chipset required) support are the things that simply cannot be measured with the typical uniprocessor benchmarks that XBITLABS are using in their review.
3. I think that AMD will use the next few months before real Athlon 64 launch to improve the things , both in CPU and their chipsets (8xxx series).
4. I was pleasantly surprised to learn that the memory controller, embedded in the CPU die, is supporting (already) DDR 400 memory.
5. All benchmarks are 32 bits, on 32 bits OS (windows). I think that XBISTLABS crew don’t made any efforts to test the almost ready Linux distributions with native x86-64 support, as we know there are many of them around.
Cheers.
Just a stupid question, but you will need a Windows XP optimised version for 64bits?
No, you can just download the appropriate libraries and source and recompile your Linux system to take advantage of it right now. I’m sure there are several distributions you can find that may already have binaries for X86-64 systems, if not at least all the patches and info necessary to make use of the hardware.
Boring, this doesn’t even equal the DEC Alpha work. Why are they going backwards? Just to keep behind Intel?
Sorry…I posted it in a too innocent way. What I want you to notice is that most of the market is based on the “wait and see” so this is gonna be a total failure for AMD, actually it’s going to be a complete disaster for them.
1.- It’s all about the software, they need the software to move to 64bits, please forget about Linux and servers…that is not the real word thas is “geek word”, 7% of the market, and what is worse, are they considering investing during this year?. Today the money for “processing” is on Video and Laptops…it seems that Nvidia and Intel know where the money is. Ok, so AMD will have a 64bits “futuristic” processor while Intel sells P4 with HT like crazy.
2.- The market they are targeting is really conservative. To make such a transition they will wait till Intel markets something similar and compare or wait till Microsoft does something radical. What if Intel doesn’t goes to 64 soon, more wait and see, no good for AMD.
The only company that may have some success “right now” on the 64bits market may be possibly Apple. Just think about it, do you really need 64bits to run Abiword now? To run a server now? If I was unprofitable and my secret plan were Opteron…wow.
It is difficult to take a ‘test lab’ very seriously when they didn’t bother to put a 64 bit OS on their test machines, but keep complaining about the possible unrealistic results in a 32 bit enviroment. They could have at least hired someone for this single occasion to install a proper OS on their puters if they are unable to do that themselves. I even think that someone would have done it for free for them (just to play with linux on AMD-64.
http://lwn.net/Articles/25416/
The chip hasn’t even been released yet, no one knows how many hardware vendors have signed on, and there are no benchmarks for enterprise applications.
That review used an old revision of an engineering sample, a 32-bit OS and 32-bit applications. I think that the chip performed pretty good, considering the situation.
We only have to wait a couple more days before the real facts and benchmarks are released to the public.
Something a lot of people are forgetting is the Opteron doesn’t have to be the fastest single processor on the market when it’s going to be so easy to link eight or more together in a system.
Also, the capability of running existing/legacy 32-bit applications and operating systems is one heck of a benefit for companies preparing to transition their software to 64-bit.
I think you’re all missing the point of the Hammer. The chip is not a dedicated 64 bit chip, like the ailing Intel Itanium which has only sold 5000 units. The key to the Hammer is that this processer runs present applications at speeds which will rival the best dedicated 32-bit chips, while allowing you to painlessly move towards 64-bit apps. These benchamarks are simply not valid, because of the beta silicon and beta bios they used. However, it doesn’t make sense to say that they’re not valid becuase they didn’t run a 64-bit OS, becuase the majority of Hammers probably won’t run ANY 64 bit code for years, becuase there’s no need for 64-bit OSes outside of the server market.
To sum up, Hammer is a 32 bit chip that can run 64 bit code for practical applications. These benchmarks are bad becuase they used engineering samples, but they were perfectly justified in using 32 bit software to test it.
Apparently you weren’t aware that Microsoft is already planning Hammer support:
http://news.com.com/2100-1001-890698.html
http://zdnet.com.com/2100-1103-890467.html
Then there is plenty of linux support already (what you call geek world) including such household non-geek names as IBM:
http://news.com.com/2100-1001-861371.html
http://www.theinquirer.net/?article=5295
http://www.infosatellite.com/news/2002/07/a310702ibm_hammer.html
It looks like hammer actually has a pretty bright future. Between optimized Windows and *nixes, everyone but BeOS fans will have something to use with Hammer.
For some reason I was thinking this was about Hammer instead of Opteron, anyways, the same support is there for Opteron as for Hammer.
Opteron and Hammer are the same. Athlon64 and ClawHammer are the same too.
on thing that differ the athlon64 from opteron is that the opteron has a dual memory controller to increas the meory bandwith two times.
i dont know if there is anything more that differ. well i dont think nither of them will be the almigty killer on 32bit apps but they dont have a bad performance on 32bit it will be nice tho se some test on a 64bit platform
You are wrong. I DO KNOW about Microsoft future support. But still you miss the point. If Opteron hits the streets in two months what will be the point in buying it instead of a 3.X Intel P4 with HT?.
The 64bits only make sense if you do a WHOLE TRANSITION to them not just; ok we are planning to support it. If not, I would rather chips with less power consumption for example.
Don’t misunderstand me, I’m just saying I don’t see sales anywhere. I see a complete useless chip “right now” away from the inertia of the market. Well you can always come back and say it was before it’s time but hey look I ran debian 64 on it, cool!. BTW what makes you think that any linux based company may move to 64bits? Maybe in 12 months? Please tell me your reasons for the “bright future” of Opteron.
BTW; IBM has it’s own 64 chip roadmap based on PowerPC I’m sure they are supporting that more than AMD chip.
Who from IBM said AMD was disappearing and only would remaind Intel and IBM?
You are wrong. I DO KNOW about Microsoft future support. But still you miss the point. If Opteron hits the streets in two months what will be the point in buying it instead of a 3.X Intel P4 with HT?.
I am sorry if I offended you, your previous post made it appear that you weren’t aware of Microsofts support for it.
The 64bits only make sense if you do a WHOLE TRANSITION to them not just; ok we are planning to support it
This is wrong. The reason that AMD went with an x86-64 instruction set instead of a new (and better) instruction set is to keep backwards compatibility with existing 32-bit applications. Migrating everything at once is expensive, both monetarily and in downtime. x86-64 allows a less abrupt and painful transition to 64-bit.
Please tell me your reasons for the “bright future” of Opteron.
Because 64-bit processing has numerous advantages over 32-bit in the area of memory addressing, content creation, etc.
For more info check out this Ars Technica article: http://www.arstechnica.com/cpu/03q1/x86-64/x86-64-1.html
If I have a use for 64-bit processing, and I don’t have to give up anything to have it, I am more likely to begin assimilating it.
64 bit solves the dreaded 4 GB memory limit. RAM is now about US$150 for a 1 GB stick and falling fast. Within a year 4 GB of RAM will probably cost as little as US$200. Bigger 2-4 GB RAM modules will hit the market within a few years.
Why would you want more than 4 GB of RAM: superfast video editing in a RAM disk, engineering and scientific workstations, photorealistic games etc.
Believe it or not there are many situations where even a current Beowulf cluster is too slow such as proteomics (the Next Big Thing). Computers aren’t just used for word processing and Photoshop.
This is a 64 bit CPU executing legacy, 32-bit code, and it’s almost as good at it as the 32-bit counterpart (Athlon XP)!!?? Intel has reasons to be worried: did you see how the Itanium executes 32-bit code? 1 GHz Itanium 2 is approximately as performant as a PII 300 MHz!
The Opteron will be, it seems, an excellent solution for the transition from 32 to 64 bit applications.
BTW what makes you think that any linux based company may move to 64bits?
Because at least two Linux companies have already announced 64-bit versions of their flavors. Mandrake is one; I don’t recall the second (SuSE, I thing…help, anyone?); RH can’t be far behind.
Noone truly knows the future and can simply give their opinion on the matter based on current “facts”. Remember the amount of people that said we’d never need 32bit? Some of them may be here today. Remember Intel’s MHZ marketing strategy? Well now like the console world the “64bit” marketing could well e successful. Any new buyer or gamer not looking for the cheapest way out will definitely consider the 64bit solution. If not for workload, cause it sounds cool. That is human nature. Now you guys know Apple and Microsoft are going to be there cause they recognize this new fad. the whole CG/film industry has been crying for a cheaper 64bit solution than Sgi Fuel and Onyx workstations. Now that may not count for a large number of sales relative to mainstream sales, but it will start to move along. I know you all want one too, so embrace the mainstream 64bit revolution.
Mandrake, and I believe we can expect one from Redhat soon enough.
RedHat announced support back in 2000
http://www.redhat.com/about/presscenter/2000/press_amd.html
The xbitlabs benchmark don’t apply to Opteron, and would hardly apply to Athlon 64 when it is released. The Inquirer article is just specullation.
@ernesto
Sorry to dissapoint you but Dell, IBM, and Sun are waiting for Opteron.
How many hundreds of millions did they lose last quarter, and the quarter before, and the quarter before?
Sorry guys if I sounded a little bit aggressive in my comments. But as a matter of fact AMD is not de leader in chip processors and that implies you may come with something spectacular to get market share.
Just some points again (sorry if I’m reiterative)
1.- Support doesn’t mean sales or adoption.
2.- It’s not going to be the fast chip when it hits the street
3.- It’s not going to be the cheap chip when it hits the street
4.- It’s not going to hit desktops or laptops in short period
5.- Intel is not going to sleep and has plenty of resources
6.- AMD is on red
Although you may tell me there are OS for 64bits, the whole 64bits jump can be only adopted by leaders, that means Intel and MS, or companies with the full packet like Apple.
There are three weak points on your argumentations
IBM; c’mon they are supporting every single little piece of software or hardware that hurts MS and they make 970 chips that are 64bits (that doesn’t sounds good for AMD), actually I’m sure they will buy AMD facilities when they crash.
Red Hat, Mandrake, Suse; c’mon they are no more than 10% market share, they are going to support whatever to get customers, if tomorrow I build my fantastic 128bits processor who is going to support it first Linux companies or MS?.
Dell?; I can’t understand this one, Dell doesn’t not offer anything they just package. I don’t see why they should wait for anything?. The just go, watch and buy whatever it’s successful. So I AMD must trust Dell they are complete dead. Dell can just stop buying Opteron in one months.