And the Magic Number is.... 3.2 GHZ
Yup, Cell PS3 has it, PowerPC-ish xbox360 has it...
Did some guy at IBM sit back and say... hmm... tonights winning number is...
3.2
Let now the speculation fly on PowerMac and PowerBook G5s nex gen chips, will it too have the magic 3.2 ghz number attached to it? okay, probably not powerbook G5 but
my point is... (i have a point...) ...the PS3 and xbox360 has just added some major fuel to the speculation on next gen Apple G5 breaking that psychological 3ghz barrier
Did some guy at IBM sit back and say... hmm... tonights winning number is...
3.2
Let now the speculation fly on PowerMac and PowerBook G5s nex gen chips, will it too have the magic 3.2 ghz number attached to it? okay, probably not powerbook G5 but
my point is... (i have a point...) ...the PS3 and xbox360 has just added some major fuel to the speculation on next gen Apple G5 breaking that psychological 3ghz barrier
Comments
Cell and Xenon are both supposed to be high clock speed, simple logic CPUs. There were designed to attain 3.5+ GHz clock rates on 90 nm fabs. If IBM is shipping both of them at only 3.2 GHz, something is still going wrong at Fishkill. Either that, or all the 3.5+ GHz clock rate advertizing was a marketecture conflation of 65 nm w/90nm.
I get the feeling those PS3s aren't being water cooled.
However, will you really be staggered by the difference with 2 x dual core 3.5GHz CPUs?
Maybe, it depends on the rest of the motherboard, the GPU, the system and what you're doing.
If you can afford one and actually need the performance, go out and buy a dual 2.7GHz or a pile of Xserves.
Do your work. Make enough money to be able to afford something better as soon as it comes out.
Originally posted by THT
I don't know about you, but Cell only shipping at 3.2 GHz is bad news for 970mp/gx shipping at 3+ GHz.
Cell and Xenon are both supposed to be high clock speed, simple logic CPUs. There were designed to attain 3.5+ GHz clock rates on 90 nm fabs. If IBM is shipping both of them at only 3.2 GHz, something is still going wrong at Fishkill. Either that, or all the 3.5+ GHz clock rate advertizing was a marketecture conflation of 65 nm w/90nm.
Nothing is "going wrong at Fishkill". Nobody else can ship a 200+ million transistor part at >3.2 GHz at a power level that can appear in a game console along with half a gig of memory and a 200+ million transistor GPU either. IBM is doing pretty damn well to pull this off (the processor design in these things is well designed to limit leakage), but it doesn't change the fact that we hit a wall at 90nm and power leakage is limiting us severely.
Dual 2.7 GHz G5 = ~41 Gflops
PS3 is 53x more powerful than Apple's top of the line G5. To be fair, the G5 is shipping and the PS3 won't be shipping until next year.
To say the PS3 is a quantum leap over everything else is putting it mildly. The PS3 is to the PM G5 as the PM G5 (dual 2.7) is to the 1995 PM 7500 (100 MHz).
If there could be a PM next year with dual Cell chips at 4 GHz, it'd trump everything else. We're talking some 5 Tflops. Sure would be dandy for 3D animation, high end visualization, HDTV editing, and scientific work.
Originally posted by Programmer
Nothing is "going wrong at Fishkill". Nobody else can ship a 200+ million transistor part at >3.2 GHz at a power level that can appear in a game console along with half a gig of memory and a 200+ million transistor GPU either. IBM is doing pretty damn well to pull this off (the processor design in these things is well designed to limit leakage), but it doesn't change the fact that we hit a wall at 90nm and power leakage is limiting us severely.
"going wrong at Fishkill" was perhaps hyperbole, but Cell was marketed in February to run at 4 GHz, not 3.2 GHz, on 90 nm with relatively low power consumption (50 to 80 Watts). Actually, I don't even remember 3.2 GHz even being in the low end of the clock range thrown about at the time. The high clock rate was a requirement for its uber-performance, no? Now at 3.2 GHz, it's lost nearly a quarter of its slated clock rate and something close to that in performance.
As far as a 200+ million transistor part at 3.2+ GHz, Intel is already doing that, shipping now, albiet in very low quantities. But that's really not that important to me, performance is what really matters. It doesn't matter how they get it (most of the time). We'll see how the Cell and Xenon performance are when the boxes are near shipping.
[edit: grammar]
Originally posted by Rolo
....
If there could be a PM next year with dual Cell chips at 4 GHz, it'd trump everything else. We're talking some 5 Tflops. Sure would be dandy for 3D animation, high end visualization, HDTV editing, and scientific work.
And people would still be saying, "But it still runs Microsoft Office slow, games still suck and there are no high end video cards available.
On an interesting side note, on a trip to a local resort to do some Mac training/consulting in the art dept., I met a fascinating person. there head of "IT' wanted to meet the "Mac Guys" personaly, so i accpected the wost as you may imagine, but it was just the opposite. this gentleman was an avid Mac user, altho his curent carier made him move much of his research to wintel, his love was Apple/PPC tech. Come to find out, as I feel out of my chair, he was on the original design team for Altevec while working for IBM for most of his adult life!!! He told me some stories about meeting Steve J, about the 970 being on the whiteboard fas far back as 7 or more years ago, and many other amazing stories. Very very nice, mellow guy. I'm so happy to know him and have his email!
Originally posted by mikenap
i smell roses for Apple with IBM's success in gaming chips. I dont believe it will stifle development of the Apple parts, but enhance there dev with better tech, more funds for R&D, etc.
i feel the same way. there are far more game consoles sold each year than personal computers. so the market for IBM's PPC tech just increased dramatically, and the profit margin on the console chips is likely much higher (volume sales per R&D dollar).
all this means more future R&D money for the PPC architecture, from which apple will benefit, even if as a small fish in the market pie.
Originally posted by imiloa
all this means more future R&D money for the PPC architecture, from which apple will benefit, even if as a small fish in the market pie.
in anticipation of the rebuttal: yes, MOT's R&D focus on their primary embedded market did leave apple dry for many years. but i expect the console focus to be different, since the console design goals are more similar to PC goals than embedded designs are.
as has been discussed in depth in other threads, cell is a specialized processor, different from 970. but where embedded speeds could be "good enough," console chips are just as hungry for speed as PC chips. so the design goals are more aligned: heat/leakage, bus speed, branch prediction, etc...
Originally posted by rickag
... And people would still be saying, "But it still runs Microsoft Office slow, ...
Has anyone else noticed that Entourage is incredibly slow accessing Exchange (via IMAP)? It's far slower than running Outlook in VPC...
Originally posted by MarcUK
does anyone know what kind of ppc is the main core in the cell? Is it equivalent to a 970, or more like an xbox360?
We've been calling it PPE or PPE-derived, which actually comes from the Cell ISSC conference presentation in February. The PPE is an acronym for the PowerPC Processing Element in the Cell. The microarchitecture of the PPE is said to be derived from a high CPU clock speed research project IBM had about 4 to 5 years ago.
We know that PPE is a 2 instruction issue, 2-way SMT PowerPC core with a 128 bit VMX unit. We do not know if the VMX unit can execute all 162 AltiVec/Velocity Engine/VMX instructions or not, but it is likely to have custom SIMD instructions per Microsoft and Sony requests.
We believe that Xenon (xBox 360 CPU) has 3 PPE cores because the Xenon cores are described to be 2 instruction issue, 2-way SMT custom PowerPC cores. That has to be the Cell PPE or something very close to it. They also both have very high clock rates (3+ GHz), which would be further evidence for Xenon having 3 PPEs.
The PPE has no known relationship to the 970 microarchitecture. The 970 is a 4 instruction + 1 branch issue architecture with a lot of instruction-ordering logic and execution units. The PPE is "dumb" but clocks really fast.
Cell has 1 PPE core and 8 SPE cores. Xenon has 3 PPE cores. They both have been announced to be at 3.2 GHz when the respective game consoles ship. At this point, we don't know if the cores are identical or slightly customized for each customer, but they definitely are in the same genus of the PowerPC Family or PowerPC Order. (Which reminds me that I need to update me PowerPC cladogram
Nintendo is also using a PowerPC CPU to power their gamesole, but we have no idea what it is yet, just that it is backwards compatible w/all Nintendo games. The Nintendo GameCube used a ~450 MHz PPC 750 CPU with a few custom instructions. So, it may be 1.6 GHz 970 for all we know.
Originally posted by Rolo
Single 3.2 GHz PS3 = 2,180 Gflops
Dual 2.7 GHz G5 = ~41 Gflops
PS3 is 53x more powerful than Apple's top of the line G5. To be fair, the G5 is shipping and the PS3 won't be shipping until next year.
Sorry people keep making this mistake. The PS3 number includes the graphics card. The cell chip delivers 218 GFLOPS.
what are the possibilities of the cell chip being used as a general purpose AltiVec DSP enhancement to the 970MP. Kind of like the FPU of old being a seperate add-on?
Could it be connected over a hypertransport bus to the main processor to allow for massive vector calculations?
Or could it be connected like a Gpu over PCI-Express but for DSP vector processing? I'm thinking custom processing for apps like FCPro, Motion, and Logic.
Originally posted by Telomar
Sorry people keep making this mistake. The PS3 number includes the graphics card. The cell chip delivers 218 GFLOPS.
thanks for clearing that up.
and thanks people for responding to my thread..!
is there a GFLOPs myth now where higher GFLOPs straight away mean better performance? or again does it depend on chip design, pipelines, etc, etc, altivec, etc? etc? i'm confused
bleah... no wonder i did software when i minored in computer science, not hardware engineering
Originally posted by MarcUK
thanks THT, so its kind of like a dumb controller?
No. The PPE is a full fledged CPU core. Cell is analogous to a system with a 2-way SMT PPC CPU (PPE), 8 vector coprocessors on a ring-topology bus to the CPU, and memory and I/O controller chips, except the whole thing is on one piece of silicon. An operating system would see the Cell as a dual processor PowerPC and 8 vector processors (I think). Not sure if the SMT can be turned off.
Likewise, and operating system should see the Xenon system as a 6 PowerPC CPU SMP machine.
what are the possibilities of the cell chip being used as a general purpose AltiVec DSP enhancement to the 970MP. Kind of like the FPU of old being a seperate add-on?
Could it be connected over a hypertransport bus to the main processor to allow for massive vector calculations?
Or could it be connected like a Gpu over PCI-Express but for DSP vector processing? I'm thinking custom processing for apps like FCPro, Motion, and Logic.
Where have you been, man? Speculation has run wild.
I think the answer is yes for all of the above. But, who knows what Apple wants to do. IBM, Sony and Toshiba want to put it into all things media (HDTV everything), controllers, and media workstations.
THT: VMX is VMX. If it doesn't support every VMX instruction, it's not VMX.
Originally posted by imiloa
in anticipation of the rebuttal: yes, MOT's R&D focus on their primary embedded market did leave apple dry for many years. but i expect the console focus to be different, since the console design goals are more similar to PC goals than embedded designs are.
as has been discussed in depth in other threads, cell is a specialized processor, different from 970. but where embedded speeds could be "good enough," console chips are just as hungry for speed as PC chips. so the design goals are more aligned: heat/leakage, bus speed, branch prediction, etc...
But the upgrade cycle of consoles is a lot longer than it is for computers, 3 1/2-4 years vs. 8-12 months. Hopefully this will give IBM a bit of breathing room to focus more on the next gen Power5 derived chip and/or a low power 970 for laptops. However if all 3 console players last through this round or releases, and they stay with PowerPC/Cell chips then IBM will have to get busy on the next gen console chips within 12-18 months.
Originally posted by THT
I don't know about you, but Cell only shipping at 3.2 GHz is bad news for 970mp/gx shipping at 3+ GHz.
Cell and Xenon are both supposed to be high clock speed, simple logic CPUs. There were designed to attain 3.5+ GHz clock rates on 90 nm fabs. If IBM is shipping both of them at only 3.2 GHz, something is still going wrong at Fishkill. Either that, or all the 3.5+ GHz clock rate advertizing was a marketecture conflation of 65 nm w/90nm.
Nr9's prediction in his "power mac won't get any faster" thread holds.