Slashdot Mirror
AMD Athlon 64 6000+ Launched And Tested
Spinnerbait writes "AMD officially launched their next speed bump in the Athlon 64 product line,
in the form of
a new 3GHz part branded the Athlon 64 6000+. This new dual-core Athlon
64 sports 1MB of on-chip cache per core and is designed for AMD's Socket AM2
platform. This chip is still built on AMD's 90nm fab node and is comprised
of some 227 million transistors. It also carries a thermal power profile
of about 125Watts. Unfortunately, in all the
benchmarks seen here, it was still unable to catch Intel's Core 2 Duo E6700
chip at 2.66GHz."
At least it uses more power!
I've always wanted to try running DOS on a processor with 1MB of L2 cache...there's just something retro wicked about running an OS where the entire base memory fits in on-die cache.
I have to wonder if qemu and the kernel's kvm will allow me to dedicate an entire core to a DOS image.
tasks(723) drafts(105) languages(484) examples(29106)
In the full announcement they also mention new 45W single-core desktop processors: Athlon 64 3500+ for $88, and 3800+ for $93.
I would still buy this over intel's processor, my god, that thing has alot of pins
WulframII - Free Online Mutiplayer 3D Tank Shooting Game
What is the point of releasing a new iteration of an existing platform to bump up speed and still not catch up with the competitions products?
Wouldn't they have been better served re-routing this R&D effort/money into something which would put them back on top of either the price or performance curves?
I'm a fiscal conservative, it's a pity we don't have a political party anymore
The problem with socialism is that they always run out of other people's money. - Margaret Thatcher
"Unfortunately, in all the benchmarks seen here, it was still unable to catch Intel's Core 2 Duo E6700 chip at 2.66GHz."
What's unfortunate about it? It's just a fact.
"Ask not what your country can do for you." --John F. Kennedy
From TFA: The OS used was Windows XP Pro SP2.
A 32 bit OS. The real strength of the AMD 64 architecture is running in 64 bit mode - benchmarking this chip compared to other 64 bit architectures would be far more helpful than running a 32bit Sandra tests and Photoshop tests on it.
Not a very helpful benchmark. I'd like to see these chips compared running 64 bit OS's - and compare the speed and throughput of applications like Apache, Oracle, PostgreSQL, MySQL, PHP / Perl scripting, and raw image processing - not Photoshop, where most of the time is spent waiting on the user to do something.
AMD has been skimping lately on its cache. I have a sneaking suspicion that the majority of AMD's current performance issues are related to cache and lack thereof.
The Intel chips carry 4 to 8 Mb of cache. The thing about the Intel architecture is that the cache is shared across both or all 4 cores. In contrast the AMD chips have a dedicated *tiny* 1 MB cache for the consumer chips and 2mb per core on the high-end parts.
With that said, the reality of dual core computing is that one core is used much more heavily than the other. In Intel's case this means that one core is basically given the entire cache for its use - a significant performance boost when running a few tasks. In AMD's case the idle cache is inaccessible to the heavily loaded core.
The reason that makes me think that the cache is the current bottleneck is that the memory controller on the AMD chip is significantly faster than Intel's. Given that fact one would conclude that in non disk-bound applications that require large amounts of memory (games) the AMD chips would pull ahead. This is not the case. Of course there is more than just cache at play here but the fact that the Intel chips has 4 to 8 times more cache available to it has to make a fairly significant difference.
Check out my AMD FX-70 at http://amd4x4.blogspot.com/
The percentage of chips able to run at a given frequency rises as they tweak the process to make manufacturing more efficient. This is not a new factory, process or design. They make them already. Why not sell them?
What? Do you mean using some sort of BigInteger-like library? Otherwise adding two big numbers is a single operation (assuming they're already in registers).
Heh, who needs a 64 bit processor to try that? Python works fine
:)
>>> 1000000000000000+1000000000000000
2000000000000000L
>>>
1 quadrillion + 1 quadrillion = 2 quadrillion. Have a nice day.
Ewige Blumenkraft.
Is AMD hoping that they can just keep increasing the multiplier until people forget that all that does is increase frequency and cause more cache page misses? Do people not remember when the processor that was increased in frequency by .2 Ghz through multiplying the clock, that it actually performed worse than it's predecessor that performed at bus speed? If AMD does not increase the bus speed they are always going to be playing catch up to Intel.
Seems silly to release another 90nm part before the move to 65nm but keep in mind their are DIFFERENT LINES/TEAMS working at AMD. It's not like the production people working on retail 90nm parts are the same as the people testing new 65nm techniques.
AMD is just trying to get as much non-idle time out of the fab as possible before they move everything to 65nm.
It's the same reason why they make "el-cheapo sempron" parts and sell them AT A LOSS. It's better to lose a few bucks than a lot. And idle time in a fab costs a lot of money.
Tom
Someday, I'll have a real sig.
Even if we paid Indians $1 an hour, at a rate of even 100 transistors per hour, this would cost $2m to make one of them and this is not counting material. How many government agencies can afford that ?
The design and development of a processor has improved vastly since the days of borked multipliers. There are standard benchmark tests that engineers use to rate their designs in-house. If AMD chooses to go with smaller caches, I would imagine they have very good reasons.
Perhaps in order to keep good performance when communicating between caches they need to keep the number of memory addresses low so that the overhead stays low. They decided that separate caches was a better model, and they currently have to maximize performance with this design.
AMD might have favored their server market when choosing this design and separate cache works better for server machines. They may need to refine their architecture for the desktop market. Don't be so quick to accuse AMD of making cache mistakes without doing the math for find the theoretical best solution.
To comprise is to be composed of. The chip comprises 227 million transistors, or is composed of 227 million transistors.
This is one of the rare cases where a common misuse isn't just a gradual development of language; it actually reverses the sense of a word, replacing the relation of the whole to the parts with the relation of the parts to the whole.
My blog: http://www.seebs.net/log/ --- My iPhone/iPad app: http://www.seebs.net/seebsfrac/
Comment removed based on user account deletion
Critical question here is whose compiler was used to compile the benchmarking programs??? and whether the benchmarks themselves were specifically compiled to take advantage of the instruction sets...
There's only one truly independent test available... and that's how long it takes to emerge a default Gentoo install with the compile options set to match the respective processors. Everything else should be as identical as possible.
Donald 'Duck' Dunn: We had a band powerful enough to turn goat piss into gasoline.
But I use Ruby you insensitive clod!
irb(main):001:0> 1000000000000000+1000000000000000
=> 2000000000000000
Guess I should get that cup of coffee now, eh?
-Ben
Good point.
Are we comparing AMD 2 cores with Intel 4 cores here?
If so then I humbly suggest that the test might just be a tad skewed.
It does.
With Athlons, the memory controller are on the northbridge (just like Intel's). You can put whatever memory on the mother board, as long you put the correct north bridge.
So that's why you have both SDR and DDR Socket A mother boards.
That's not how is works with Athlon 64s : They have on die memory controller. The type of memory you can connect to the mother board is directly determined by the type of processor. And until recently each type of processor has it's own connector :
Single Channel DDR : Socket 754
Dual Channel DDR : Socket 939
Dual Channel DDR2 : Socket AM2
Only from now on will you have mutually compatible AM2/AM2+/AM3 mother board, which will use mechanically compatible connectors and the limitation will be only be the memory controllers on the chips (AM3 processors have both DDR2 and DDR3 and can got in all 3 motherboard. AM2 chips only have DDR2 and only go in AM2/2+ MB).
On Athlon 64 motherboard, the nortbridge is nothing more than a controller in charge of peripherals and their busses and doesn't touch the memory at all. It's completly agnostic of the memory and only speaks "Hypertransport" to the CPU. It is mutually interchangeable with all mother board. And in fact you can find the exact same VIA KT880 AGP chipset on mother board from 754 all the way up to AM2, regardless of the memory.
You can make different king of motherboard with the same chipset.
But 939 Processor can only connect to DDR memory, so you're stuck with it.
(On the otherhand, we could imagine building PCI-e nForce6 motherboards for Socket 754 CPUs and AGP KT880 mother board for AM3 connectors. But no company curently bothers.)
As a side note, that's one of the reason why Athlon64 have a smaller cache :
- Unlike Intels they're not limited by the bus speed for memory transfers. They have access to memory at full speed.
- Memory access is direct, without having first to be processed by north bridge and latency is much lower.
Of course now that DDR2 (and even more DDR3) have higher latency, these advantages don't shine any more.
To see it by yourself can look at the trace on the mother board. On regular mother board, the north bridge is in the middle and has trace both to the memory and to the CPU. The CPU is only linked to the northbridge.
On athlon64 mother board, the traces go from the memory to the CPU. The north bridge is only connected to the CPU.
In fact now that the AM2/2+/3 familiy has been declared upward compatible, you may start to see the same kind of compatibility that we had back with the Slot-1 connector which could be used with the first Pentium IIs all the way up to the latest Pentium II Tualatins (given one uses the correct slotket).
And this what exactly this is all about : AMD *does want* a stable socket so they can attract potential chip makers that will be interested in making specialized coprocessors that will remain compatible thru all upgrades from AM2 to AM3.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
- Athlon 64 X2 is their dual-core offering, available for Socket AM2 and Socket 939 motherboards.
- Athlon 64 FX is their high-end single-core offering, available for Socket AM2, Socket 939, Socket 940, and Socket F (server) motherboards.
- Athlon 64 is their mid-range single-core offering, available for Socket AM2, Socket 939, and Socket 754 motherboards.
- Sempron is their low-end ("value") single-core offering, available for Socket AM2 and Socket 754 motherboards.
So, to answer your questions: Athlon FX's are not dual core (dual core chips will be labeled Athlon 64 X2), no matter what socket they use. Intel does have four-core offerings, and they do call them Core 2 Quads. The Core 2 Extreme is not necessarily a four-core chip; there are Core 2 Extreme Duo and Core 2 Extreme Quad flavors.Read my blog.
AMD still does maintain a stronger memory architecture, stream numbers out of AMD platform systems are much higher than comparable intel.
Of course, this processor does nothing to widen that gap or anything..
XML is like violence. If it doesn't solve the problem, use more.
Wrong.
All of the Athlon 64 FXs newer than the FX-60 are dual-core. This includes the FX-60, FX-62, FX-70, FX-72 and the new FX-74.
You are correct, there are dual-core Athlon 64 FX's. My mistake. I blame AMD's web site, which doesn't make this immediately clear on the product pages.
Read my blog.
I'm still waiting for the next big thing from AMD. I really don't have a need to upgrade at this point, my gaming rig is a 2.4ghz X2 and as far as I can tell, it does everything I throw at it fast and smooth. Would I like a faster CPU ? Of course! But I don't think a 20-25% clock speed boost would excite me enough to justify the expense. I have high hopes for the quad-core Barcelona coming later this year, which will be a native quad unlike the Intel Kentsfield which is just two dual-core dies smushed together on a package. I can only hope it will be as satisfying of a step forward as the Athlon64 -> X2 hop was when it came out.
Why do I resist Intel ? Mainly cost. I've always preferred the chipset panorama available to AMD platforms. Maybe that's because I've seen too many pricey Intel boards catch fire on my test bench, but those were old P4 setups. I haven't laid hands on a Core 2 board yet, I can only hope they're less tempermental than their ancestors. The price premium still remains, as you can get a perfectly decent AMD board for about $90 canadian ($75-80 USD), but Intel boards under $99 tend to suck hard, if you can find them at all. The mean price for an enthusiast Intel board seems to be $200 and up, double what a like-featured AMD board would be.
Add in the CPU price differences and you can easily build up a 20-25% price premium for the Intel system, versus a $1200 mid-range AMD rig. Sure, the Intel is faster now but a lot of people would rather put that extra $200-300 toward a video card, more RAM or a pair of Raptors. It's all too easy to forget that we're building a System, of which the CPU is just one part.
I remember this kid came to me, a few years ago. He had cobbled together a crapwagon of a PC, with a bunch of hard drives spanning an entire decade.. 3.2 gb all the way up to 250gb, extra IDE controllers to run them all. Then he had this 420-watt no-name power supply, PCI video card because he already fried the AGP slot! But then he had an overclocked P4 with watercooling and the works. Well that day I had just received a batch of new cpus and sure enough, he dropped $600 on the fastest one I had. I wanted so bad to give him the next one down and have him spend the $200 or so difference to replace the other garbage in his case, but his only priority was achieving the highest clock speed. He didn't even do anything intensive with his PC, all he cared was to have the fastest benching CPU for screenshots and bragging rights. The fact that it could do high-speed fuckall didn't matter one bit to this teen.
-Billco, Fnarg.com
"Unfortunately, in all the benchmarks seen here, it was still unable to catch Intel's Core 2 Duo E6700 chip at 2.66GHz"
No shit, Sherlock. Core Duo is 32-bit and can only run 32-bit XP or Vista. When you limit a 64-bit processor to a 32-bit OS, of course you're most likely gonna get your ass nailed to the wall.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
Why exactly do you find that the given benchmarks are not useful? Do you think that we should care about how the chip could possibly perform in a scenario that accounts for a tiny proportion of all real use(your 64-bit suggestion) or that perhaps we should care about the use that these chips will see the vast majority of the time(32-bit apps on a 32-bit OS)?
I think somehow that the latter is more useful.
You can tell the Intel chips apart by their model numbers. Extreme edition means the multiplier is unlocked both up and down. All the rest have locked multipliers that can only be decreased.
Extreme edition (X, QX):
X6800 is dual-core.
QX6700 is quad-core.
Also Quad, not extreme edition (Q):
Q6600
All the rest are dual-core (E), Conroe (E6):
E6700 4 MB cache
E6600
E6400 2 MB cache
E6300
Allendale (E4):
E4300 2 MB cache
Not if the Intel CPU's are still price competitive, because that's all that matters to your end user.
But no, these Core 2 Duo's aren't quad processors -- in that case they'd be talking of QX6800's, not X6800's.
Beware: In C++, your friends can see your privates!
You're off by a couple months. The high end Athlon FX processors have been dual core for a while now.
http://www.amdcompare.com/us-en/desktop/details.as px?opn=ADAFX70GAA6DI
s px?opn=ADAFX62IAA6CS
http://www.amdcompare.com/us-en/desktop/details.a
Note how it says "1MB x2" for L2 cache.
-- The act of censorship is always worse than whatever is being censored. Always.
The only conclusion you can see from these benchmarks is that quad is better than dual is better than single. Between different CPU's of identical number of cores, the difference is negligible. They're all limited by memory bandwidth.
my hypothesis is that addition breaks down when you get into numbers bigger then about a trillion, in such a way that the result of adding two numbers tends towards the result of multiplication
Interesting. My hypothesis is that multiplication of two such numbers will differ from addition of two such numbers by at least 12 orders of magnitude. Let us know what you find out.
Education is a better safeguard of liberty than a standing army.
Edward Everett (1794 - 1865)
The URL got corrupted, so I was referring to this and this.
Violence is the last refuge of the incompetent. Polar Scope Align for iOS
Oh crap. It says late 18th century, not late 1800s. That's over 200 years of valid use.