AtenRa
Lifer
- Feb 2, 2009
- 14,003
- 3,361
- 136
Oh, the irony when a 6.5 year old Intel CPU is trading blows with AMD's current most advanced CPU available today.
Carrizo is the most advanced AMD CPU available today, not Vishera.
Oh, the irony when a 6.5 year old Intel CPU is trading blows with AMD's current most advanced CPU available today.
Carrizo is the most advanced AMD CPU available today, not Vishera.
Carrizo is the most advanced AMD CPU available today, not Vishera.
Carrizo certainly isn't going to be faster than an 8350.
Dave's point stands.
So Haswell is more advanced than Broadwell? I thought not. This is an academic debate forum. False dichotomies have no place here.
Obviously we're discussing products that are actually launched and available to buy. Broadwell is available to buy today. Carrizo is not.
So Haswell is more advanced than Broadwell? I thought not. This is an academic debate forum. False dichotomies have no place here.
You asked whether or not carrizo beats an 8350 knowing full well carrizo is a mobile only processor. Hence I asked whether or not broad well(also a mobile only processor) is less advanced than haswell a desktop processor too.What are you talking about? I never mentioned either of those.
Where can I buy a Carrizo CPU today? Newegg?
Looks at your post history...
Ah that explains it.
It's not about availability it was about him comparing a mobile only architecture to a desktop processor and using that as grounds to claim one processor is more advanced than the other.Still stands that one cannot buy carrizo now and there are no tests from independent sources, only claims from AMD, so any "comparison" is meaningless.
We have heard this excuse for Bulldozer (and their APUs/HSA as well) and its derivatives from day one, that it was (is) the fault of the software that performance is sub-par. OTOH, one could just as easily argue that part of the job of a hardware supplier is to design hardware that performs optimally with the software available. Admittedly, it is a somewhat of a chicken vs the egg argument, but especially when you have the second place market share it does not seem logical to expect the software to adapt to your architecture. In any case, the continual declining market share of AMD in cpus speaks for itself.
IBM is quitting the foundry business because:
1) They didn't have enough customers to sustain their business
2) POWER is bleeding share to Intel at a very fast rate.
Basically their business collapsed, and they *paid* Globalfoundries to get rid of it. I think this has to do with decommissioning costs, which would be far bigger than to pay 1 billion to Globalfoundries to take over their fabs.
I'm not telling that you are cherry picking, but that AMD processors have very irregular performance when compared to Intel processors, trading blows in a few cases, losing badly in others, and because of that the 40% number should be taken with healthy quantities of salt. The 40% can be on these cases where AMD currently falls badly, while the general improvement be much lower than this (hence my comparison with Nehalen). I believe that Nehalen IPC with Nehalen clocks with a high core count can give them a viable server product for the bottom of the stack, something they lacked with the CMT chips.
Carrizo is the most advanced AMD CPU available today, not Vishera.
If speed is the only criterion, then you're right. Then at least you would never complain about the 8350's power consumption.Carrizo certainly isn't going to be faster than an 8350.
Dave's point stands.
If speed is the only criterion, then you're right. Then at least you would never complain about the 8350's power consumption.
Regardless, GF now owns that node plus some skilled engineers who will (hopefully) put all of that skill to good work on future products.
It will be interesting to see how POWER develops in the future. You speak as though POWER is officially dead, and that nobody is selling it anymore.
I've said it before, but I'll say it again - AMD is taking plays out of Intel's playbook, albeit from a few seasons ago. It looks to me like they are trying to re-engineer Haswell on 14nm, or something very similar anyway. In all probability, they will wind up with a chip that falls somewhere inbetween Ivy Bridge and Haswell, with lower clockspeeds and moar coars to compensate.
What if speed is just one criteria?
As you can see, speed was the only criteria for carrizo to be more advanced than piledriver for Phynaz. Hence the collective backlash.phynaz said:Carrizo certainly isn't going to be faster than an 8350.
Dave's point stands.
As you can see, speed was the only criteria for carrizo to be more advanced than piledriver for Phynaz. Hence the collective backlash.
And he wouldn't be wrong. We are not seeing server Piledriver users or mid-market desktop users switching to Carrizo, so for some markets Piledriver is AMD most advanced architecture.
That is a terrible way to compare architectures. If that was the case, you could say Haswell is more advanced than broadwell in some areas because broadwell is mobile only which anyone would agree is incorrect.
That is a terrible way to compare architectures. If that was the case, you could say Haswell is more advanced than broadwell in some areas because broadwell is mobile only which anyone would agree is incorrect.
As you can see, speed was the only criteria for carrizo to be more advanced than piledriver for Phynaz. Hence the collective backlash.
What collective backlash? It's just you that doesn't seem to be getting what Dave wrote.
IBM nodes weren't really rocking the world, really. IBM wasn't able to catch up Intel in the first place, but they also failed to develop a node competitive with the 1st tier open foundries (TSMC and Samsung), so basically whatever IBM was developing and selling on the market in terms of node IP wasn't exactly bleeding edge for the market needs, both in terms of performance and TTM, to the point that the most advanced SOI implementation for SoCs doesn't come from IBM but from STM.
This team GLF acquired from IBM might be better than the GLF team that screwed up 28nm, 20nm and 14XM, but I would have doubts on whether this team is able to go head to head against TSMC and Samsung.
Death =! irrelevancy. HP kept Itanium on life support for years, and their Integrity business is nothing like the IBM Mainframe business, so I expect POWER to last a *very* long time, but less and less relevant as the time passes.
Although the POWER8 is still a power gobbling monster, just like its older brother the POWER7, there is no denying that IBM has made enormous progress. Few people will be surprised that IBM's much more expensive enterprise systems beat Intel based offerings in the some high-end benchmarks like SAP's. But the fact that 24 POWER8 cores in a relatively reasonably priced IBM POWER8 server can beat 36 Intel Haswell cores by a considerable margin is new.
It is also interesting that our own integer benchmarking shows that the POWER8 core is capable of keeping up with Intel's best core at the same clockspeed (3.3-3.4 GHz). Well, at least as long as you feed it enough threads in IPC unfriendly code. But that last sentence is the exact description of many server workloads. It also means that the SAP benchmark is not an exception: the IBM POWER8 is definitely not the best CPU to run Crysis (not enough threads) but it is without a doubt a dangerous competitor for Xeon E7 when given enough threads to fill up the CPU.
Right now the threat to Intel is not dire, IBM still asks way too much for its best POWER8 systems and the Xeons have a much better performance-per-watt ratio. But once the OpenPOWER fondation partners start offering server solutions, there is a good chance that Intel will receive some very significant performance-per-dollar competition in the server market.
It looks to me that they might be trying to reengine Haswell with lower clockspeeds and moar cores, but with much less resources than Intel had at the time. I'm particularly curious on what they will do to replace that old crossbar in the server SKUS.