Slashdot Mirror
AMD's Triple-Core Phenom X3 Processor Launched
MojoKid writes "AMD officially launched their triple-core processor offering today with the
introduction of the Phenom X3 8750. When AMD first announced plans to introduce tri-core processors
late last year, reaction to the news was mixed. Some felt that AMD was simply planning to pass off partially functional Phenom
X4 quad-core processors as triple-core products, making lemonade from lemons if you will. Others thought it was a good way for AMD to increase bottom line profits, getting more usable die from a wafer and mitigating yield loss. This is an age-old strategy in the semiconductor space and after all, the graphics guys have been selling GPUs with non-functional units for years. This full
performance review and
evaluation of the new AMD Phenom X3 8750 Tri-Core processor shows the CPU
scales well in a number of standard application benchmarks, in addition to
dropping in at a relatively competitive price point."
3 cores sounds "wrong" (it should be apower of 2, right?), but with 3 cores, you can connect each core to every other one on an internal bus much more easily than with 4 cores, since you need fewer busses, and they do not need to cross.
SJW n. One who posts facts.
I was expecting 2, 4, 8, etc. ... not 3 ?!?
Please someone explain why cores have to be a power of 2. Seriously, I can understand RAM, Hard Drives, etc. because it can easily store addresses for storage as 0x???? in hex, but for CPU cores? Why?
For what it's worth, TR reached very different conclusions after more extensive testing against more relevant competition--Intel's 45nm chips, like the Core 2 Duo E7200, E8400, and Q9300.
http://techreport.com/articles.x/14606
Too bad it's not a ternary processor as well, that would be quite an interesting product.
The idea of reviving quad cores with 1 bad core is nice, but AMD is also playing a dangerous game. It is only in AMD's interest to sell triple core CPUs when the only alternative would be to throw the (large and expensive) die out since it can't work as a quad core. However, if these things became too popular AMD would be faced with the situation of either starving the market, or taking quad cores that actually DO work and intentionally blowing the fuses to make them triple cores.
I think this might explain the pretty lackluster clockspeeds. Phenom has never clocked well, but when you can buy a 2.5Ghz quad core for not much more than the top of the line 2.4Ghz triple core, it's pretty clear AMD wants to unload these things, but not to make any big waves about it. If anything the triple cores ought to clock much higher and have substantially better power usage... but that is not the case.
AntiFA: An abbreviation for Anti First Amendment.
Looks like AMD's marketing and sales dept isn't being very smart here, pricing them the way they are. X3 chips are $20 cheaper than X4, and $5 cheaper than 2.2 GHz X4s. And with those benchmarks they are definitely not competitive against intel's 2-core and 4-core offerings. Come on guys! If you don't let go of some of the margins and price them aggressively against Intel you're going to die.
...that makes AMD more competitive and sell more processors is a good thing in my book.
After all, healthy competition keeps them honest, eh?
No sig for the moment.
Surely Intel's chips have failed cores sometimes too. What do they do with theirs? Just chuck them out? They should be reselling their failed quad-cores. Interesting: What happens if they don't have enough failed quad-cores to meet demand of tri-cores? Would they purposely disable a core that I could re-enable myself just to keep up with demand?
TFA specifies an AM2+ socket. Would I lose any functionality if I swapped out my Athlon X2 for one of these babies (on my AM2 Mboard)?
Moderation: +1 pwnage
Isn't the word "competitive" always relative? My real gripe is actually that the actual price point isn't mentioned in the blurb. I am not new enough to Slashdot to ever RTFA, so I rely solely on the misinformation in the blurb and comments.
A couple more reviews that aren't as, um, positive:
http://www.pcper.com/article.php?type=expert&aid=550&pid=2
http://www.techreport.com/articles.x/14606
http://www.bit-tech.net/hardware/2008/04/23/amd_phenom_x3_8750/1
Despite all these multiple core CPUs and, high speed I/O devices and 4D accelerating graphic cards, I still get stuck into RAM, bus or DMA bottlenecks.
Wouldn't it be better to spend some research resources into a new PC architecture with things like crossbars in order to really exploit all those parallel CPU cycles?
Maybe Computers will never be as intelligent as Humans.
For sure they won't ever become so stupid. [VR-1988]
The Tech Report has their usual in-depth coverage here: link
I would mod that insightful. lol
Sorry about the writing. Robot fingers, you know? Cliff Steele in DOOM PATROL #23
Is AMD having fab problems?
There are real 3-CPU parts. The XBox 360 has one; three PowerPC CPUs share a cache. The chip layout is four quadrants, three with CPUs and one with the L2 cache.
Is it just me, or looking at those benchmarks was the clear response to just buy intel since it wins in virtually every category anyway. Or were the intel chips listed not directly comparable? I'm still running my X2-4600+ and am thrilled with the performance... but if I were in the market, those particular charts would all be leading me to the Intel processors.
I agree, if they were smart they would have called it the "Trinity" chip, stuck a cross logo on the box, and sold it to the same Christian Fundamentalists who read the Lost Behind novels.
A failed core goes from being a sign of bad engineering, to a sign from God.
Stolen from the techreport article you posted:
'I can't help but think this all must have looked different on AMD's roadmap when it was first being put together. I doubt they expected that the fastest Phenom would only run at 2.4GHz and, in doing so, would only just match the Core 2 Quad Q6600--an older product on the way out, replaced by the Core 2 Quad Q9300. That's the reality, though, and it's constrained AMD's pricing so much that the top Phenom quad core is $235. The compression through the rest of the lineup makes the triple-core value proposition suspect. Give up a core to get 200MHz more at $195? Not likely when the Phenom X4 9850 Black Edition, at 2.5GHz with an unlocked multiplier, is 40 bucks more. The logic of the pricing scheme may be internally consistent, but the stakes are too low. I'd go with the X4 9850 ten times out of ten. If, that is, I were somehow bound and determined to choose an AMD processor over one of Intel's current offerings.'
That sums it up pretty well.
First of all, that AMD can only play in the low end of the market, and second that who is going to give up a core to save $40?
This seems like an exercise in futility.
http://lkml.org/lkml/2005/8/20/95
Somewhere in my office, I have a vintage system based on an old 486SX, with the disabled/broken math coprocessor. Who here remembers those things? Anyone? Bueller? Bueller?
I also have a couple laptops with the fully functional coprocessors. They are early tablet PCs with b/w pen-sensitive screens, and actually can do handwriting recognition with a 486DX running at a screaming 25 mhz. I might go downstairs and fire one up just for the nostalgia of it. Last I checked, they still worked.
I prefer rogues to imbeciles because they sometimes take a rest.
Who cares? Even if the chip was a failed quad core with one of the cores disabled, why is it bad for AMD to sell them as triple cores? Would you prefer they just melt the silicon back down, wasting time, money, and most importantly, energy? I certainly don't.
It would sound to me like it would run a heck of a lot colder than with 4. I mean it's designed to run at a decent temp with 4 cores running so with 3, it'll be really cold! If you underclock a processor to 75% it barely puts off any heat. Of course the 3 cores will still be maxing so it's different but it should be way cooler anyway. But of course that's a bigger problem than they think. I dunno how they're actually arranged but if 3 corners are hot and one not, plus the fact that it was a bad processor in the first place, these things are gonna fail so fast people are gonna be pissed! You don't heat a damaged straight from the factory chip unevenly!
Google's Super Secret Search Algorithm: SELECT @search_results FROM internet WHERE @search_results = 'good'
PS3 uses the CELL processor built with 8 cores and one is disabled, leaving you with 7 cores-one for the OS and 6 for games/apps. And it will boot and run a linux image, yellowdog, which is a ported centos. So there ya go, you can buy one if you want one. There's more exact specs at the links, that is a basic and probably sort of flawed summary.
I was expecting 2, 4, 8, etc. ... not 3 ?!?
Don't look at it from a marketing perspective, look at it from a manufacturing perspective. It is not a 3, it is a 4 - 1. A quad core with one broken core.
To AMD fanboi's who are reading, take a breath and do not interpret the above as an attack on AMD. This is a perfectly reasonable thing to do, why waste the three good cores and all the energy, time, and resources that went into producing them. Disable the failed core and sell the part as a trio at a discount relative to the quad.
I'm having flashbacks to the original Pentium, where a production line manufactured 120 MHz CPUs but when packaged the CPUs could be 75, 90, or 120 MHz. Some 75s were CPUs that failed at 120 and 90 but passed at 75, but many were good 120s that shipped as 75s because all the 120 orders were filled and 75 orders were pending. Hence the legendary overclocking of the 75. I wonder if dual cores will someday follow a similar pattern. The production line manufactures quads but they are packaged as quads or duos depending on testing and orders to be filled.
Please someone explain why cores have to be a power of 2.
I'm a software guy who is guessing, but I expect that it has something to do with the density of circuits on a manufacturing wafer. Square or rectangular layouts may be more natural than other geometrically tight fitting shapes such as triangles or hexagons. If so, powers of two help preserve that geometry. A linear geometry, adding each core in a line, would technically preserve a rectangular geometry but the length of the "wires" is inefficient. Keeping things as square as possible probably optimized the tradeoff between length and layout.
Because I see a lot more superstitious twaddle about "power of two" and a lot less discussion of practical performance implications.
A failed core goes from being a sign of bad engineering, to a sign from God.
That would be manufacturing not engineering, and no one gets 100% yields out of manufacturing. Not even God, look at the defect rate in his creation, human beings.
I agree, if they were smart they would have called it the "Trinity" chip, stuck a cross logo on the box, and sold it to the same Christian Fundamentalists who read the Lost Behind novels. A failed core goes from being a sign of bad engineering, to a sign from God.
Which god, Jehovah (old testament) or Neo (The Matrix)? Matrix fanbois would probably be a more lucrative market. Use the name Trinity but make the CPU packaging a glossy black instead of matte black.
Comment removed based on user account deletion
Didn't the Pentium IV teach you this? Faster clocks just burn up unnecessary power. Better to speed up with parallelizing circuitry that you can turn off when not in use. Note in the reviews that their slower chip outbenchmarked faster chips from Intel.
Dangerous game? Prices are tweaked up and down to stimulate demand for various products. When you have to compete, you have to make tough decisions about products. AMD has been in business for a long time competing against bigger established chipmakers and they will continue to be a scrappy innovative company.
Anything that makes AMD more competitive and sell more processors is a good thing in my book. After all, healthy competition keeps them honest, eh?
And it is a greener strategy, less waste of resources and energy, so there are public relations and marketing benefits as well.
What idiot modded the truth down?
I mostly do scientific programming, heavy on the number crunching, light on the data transfer. I'll take the Core 2 Quad based on what I've seen.
not a great comparison I felt. they used DDR2 memory on the AMD and DDR3 on the intel. DDR3 ram is so much more costly, that I'd think anyone considering AMD would be comparing against a DDR2 based intel motherboard.
You obviously did not read the Tech Report's review then. Yes they mention the 'oddness' of 3 cores, but their tests are quite thorough and they do discuss performace ramifications as well as all the other things you'd expect.
Why not 11 cores?
When the Raptures arrived, they surrounded and took the tree-house, took my inflatable doll and used it to bate my nerd enemies from across the street.
...
When the grass Grue too high, they hunted in Pax with the tree-house as a base; preventing me from resetting the WiFi hub up there on its solar-powered UPS setup.
I think of what awaits my friend Johnny when they took him to the Island...
At some University up in Alaska, when the Russians hadn't sold it to Abraham Lincoln yet, I heard they put people in cages to keep them from being eaten by Polar Bears.
But how did the Raptures learn to transport People and Polar Bears? I can answer that: because these are none other than Christian Raptures...they're doing to us what we did to them...but just why, oh why. did. they. want. us to have...hairy-homosexual-man sex?
Kahngress!
Kahngress!
And AMD is still shooting blanks at the year old Intel Q6600 which has hit the $200 mark on various deal sites. Even the AMD quads have their hands full against the old Kentsfield Q6600.
The beauty of it (from an engineering point of view) is that every core has been designed with 3 HT links. One goes to the memory, and two connect to other cores. So really, in a four-core system, there is an additional latency because information needs two hops to reach all of the cores. Three cores is the max AMD can do while still keeping latency at its lowest.
AMD's cores (the compute engines inside a single chip package) are NOT connected by HT links. HT links are used for communication with devices OUTSIDE of the chip package, and run at a clockspeed much less than that of the core clock.
AMD's cores are connected by a full speed crossbar switch, much, MUCH faster than HT. Most people really don't get that HT is chip-to-chip or chip-to-chipset, and that AMD has a fullspeed crossbar in the die. To say it one more time: AMD's cores within the same chip are connected at full CPU speed, and every core is exactly two hops to another: core-to-switch-to-core.
I am a viral sig. Please help me spread.
Companies price mid tier products in a way to protect the price of more expensive products. With X4 out for a while, the temptation is to drop the price on them in order to drive customer interest. By introducing an X3 at a slightly cheaper price, the buying mentality changes.
Customers who look at an X3 and think "oh an X4 is only $5 more" will buy an X4 at the higher price. No need to drop the price.
Buyers who really need to save money and go with the X3, but AMD gets near X4 revenue for them, so they don't have to write off the whole chip as a loss when the yield is off.
Always be suspicious when the midtier product is introduced after the high end product is already out - you usually end up overpaying for either one.
It's better than bad; it's good!
DRM: Terminator crops for your mind!
i believe instead they disable a not-quite-functional core from their quad-processor reject bin.
Ah, good old intel trick.
Back in the day, the 486 had a built in FPU (maths co-processor) which was expensive. The 486 could execute integer instructions about twice as fast at the same clock speed as the 386 (which didn't have a maths co-processor built in).
So, to compete with Apple, Atari (Falcon) and Acorn (Archimedes), intel launched the 486SX, which was a 486 with the broken maths co-processor disabled.
Now, there was a 386SX. The 386 was 32-bit internally and externally. The 386SX (1988?) was hobbled to have a 16-bit internal data bus and 24-bit address bus externally much like the Motorola 68000 from about 1981 (in Macs, Ataris, Amigas etc.) No maths on board.
So this is just business. "Nothing to see here. Move along," as it were.
Oh, and I can still get a proper quad-core AMD cheaper than intel's Frankenstein offering of two dual cores sewn together, so who cares?
Stick Men
Young whipper-snapper. I had a 1st generation RX-7 with a Wankel engine. 120+ horsepower from a 1146cc engine.
Stick Men
Who the fuck is modding this down ? The top AMD X3 just barely pulls ahead of a much older Intel dual. How is that a win ? Need I mention that said Intel chip is priced about the same as the AMD X3. wtf ?
Metamods: do your worst!
Dell is offering an Intel Core 2 Quad in their Optiplex 755. Boy are they gonna be mad when they find out they've been duped (or quadded?) by Intel. Maybe they'll start selling AMD processors in their desktops or something.
Invenio via vel creo
The performance review of the tri-core from AMD really shows how all of AMD's processors perform badly compared to the Intel line of processors. Every aspect AMD used to claim first in is now Intel's strength, and by a large margin in many cases. Makes you wonder what AMD is doing and has me reluctant to buy anything from AMD because they might be gone just like Cyrix.
The XBox 360 processor has 3 cores.
No one complaining.
Playstation 3 has 7 active subprocessors on its die.
Plus, there is some evidence that the 3 way configuration is faster than 4 provided they have direct connected the cores.....
This is important to support correctly hyperthreading in multiple processors configuration. Say you have a dual processor machine with each processor supporting 2 threads thank to HT. (for example a dual Pentium4 machine). If the system has only 2 application to run, the OS should prefere running those two process each on a separate processor thus getting 2x the performance, instead of running both on a single chips, one on the CPU and on the virtual processor, while the other process is staying idle, in which case the process won't get 2x the performance. Only 1x +/- something, depending on whether the HT is efficient at maintaining the CPU occupied with the other thread while the main thread is stalled waiting for a cache miss (you could get a slight performance increase), or whether maintaining two threads in the pipeline slows things down (you get a slight performance decrease).
A good NUMA-capable kernel should determine the optimal thread configuration to get the full 2x speed boost in those situation (recent Linux kernels do).
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]