If FX is designed for the future

[Vesku]

they will have to sell it in that future, ugh. A cheap Phenom II x6 is looking like a good idea now for my AMD board. That or I need to run one of those compile everything for your uArch Linux distros.

 

[toyota]

by the time that future gets here we will have even faster cpus from Intel. just think how bad Ivy Bridge is going to wipe the floor with them from a performance per watt perspective.

 

[Vesku]

Yeah, Piledriver GOALS put it at what I thought the subpar but usable FX would be. Instead it's 10-15% less IPC than my squeak by window.

 

[Voo]

That or I need to run one of those compile everything for your uArch Linux distros.

Which wouldn't help at all, since the OS itself is hardly the bottleneck and most stuff there isn't vectorizable at all (and using FP in kernels does have its own problems, cf linux kernel). Well the new registers need OS support, but afaik the current OSes support those anyhow and again just recompiling the source wouldn't help there.

Also just recompiling applications even those that could use AVX won't necessarily bring any improvement - does any compiler already produce these instructions from auto-vectorization? So basically the only improvement would be for people who update their code - in which case you'd assume they'd release the new version anyhow.

 

[Vesku]

But it also means compiling each application for your uArch which should see improvements. Faster than waiting for software makers to release a tuned version if they care to do so at all. I'd assume AMD is already working on GCC bulldozer optimizations, if not big D:

 

[mosox]

The new Bulldozer is very good in some multithreaded tasks and with more well threaded apps that will show up it will gain value and that's the benefit of an architecture that's ahead of the current software.

For instance in Cinebench R10 the 8150 is slower than the 2500K, in R11.5 (named by Anand "modern multithreaded workload") is faster.

This means that Intel is more "bottlenecked" than the Bulldozer for future applications in which the Bulldozer's new architecture will be able to gain much more than Intel. Redo this review in a year or so and we might see very different results.

As for the single threaded applications I think any modern CPU can cope with those very well, no need to sweat over that. Name one single threaded mainstream application that leaves you dissatisfied with your Phenom II, i-5 750 or even a C2Q@3.0GHz.

For gaming once one has a decent video card and is gaming at 1080p again no reason to panic. As Techspot wrote:

Finally, when it came time to play games the FX range was really no better than the Phenom II. To be completely honest, gaming on these high-end processors is so similar it's hardly worth worrying about. The FX-8150 was never more than a few frames per second slower than the Core i7-2600K at 1920x1200.

For the regular PC users the problem is not Intel but the better bang for the buck AMD Phenom II X4 and X6.

 

[zlejedi]

This means that Intel is more "bottlenecked" than the Bulldozer for future applications in which the Bulldozer's new architecture will be able to gain much more than Intel. Redo this review in a year or so and we might see very different results.

1. Intel can go through brute force aproach and increase clock speeds on Sandy Bridge by 100-200 mhz and possibly more whenever they feel like doing so.
2. Intel can move cpus with enabled HT into i5 price range and send i5 into lower market segment
3. In a year those future applications will be running on Ivy Bridge which will increase IPC, clockspeeds and drop power consumption of intel chips.

Meanwhile SB cpus will run every task with at least good performance, offer clear upgrade path to IB in 2012 and their owners don't have to pray for optimisation of every program they use to get full benefits.

 

[pantsaregood]

3. In a year those future applications will be running on Ivy Bridge which will increase IPC, clockspeeds and drop power consumption of intel chips.

Ivy Bridge will have an extremely negligible effect on IPC. It is a die shrink.

 

[Rifter]

by the time that future gets here we will have even faster cpus from Intel. just think how bad Ivy Bridge is going to wipe the floor with them from a performance per watt perspective.

Im not even worried about Ivy, im more worried about whats going to happen when Haswell hits.

AMD is beyond help for recovering before Ivy but if they dont get there crap together before Haswell then Haswell very well could be the deathblow to AMD.

 

[zlejedi]

Ivy Bridge will have an extremely negligible effect on IPC. It is a die shrink.

Yes but with every die shrink you get 3-4% higher IPC too due to various optimizations or at least higher amount of cache.

 

[s44]

Haswell very well could be the deathblow to AMD.

Intel will never deliver a "death blow" to AMD. Too many regulatory problems if that happens. We'll instead see them jack up next-gen prices or delay something or... well, a lot of things, none of which are good for us consumers.

 

[Rifter]

Intel will never deliver a "death blow" to AMD. Too many regulatory problems if that happens. We'll instead see them jack up next-gen prices or delay something or... well, a lot of things, none of which are good for us consumers.

Probably, but they cant stop AMD from dealing a deathblow to themselves, which a couple of more BD launches would do it.

 

[exar333]

Yes but with every die shrink you get 3-4% higher IPC too due to various optimizations or at least higher amount of cache.

At least IPC will not go down like BD!

 

[grkM3]

The new Bulldozer is very good in some multithreaded tasks and with more well threaded apps that will show up it will gain value and that's the benefit of an architecture that's ahead of the current software.

For instance in Cinebench R10 the 8150 is slower than the 2500K, in R11.5 (named by Anand "modern multithreaded workload") is faster.

This means that Intel is more "bottlenecked" than the Bulldozer for future applications in which the Bulldozer's new architecture will be able to gain much more than Intel. Redo this review in a year or so and we might see very different results.

so intel is bottlenecked because an 8 threaded cpu beat a 4 threaded cpu? or came close to it matching it.You do know that the 2500k has no hyperthreading right?

heck just wait until a real 8 core sandy E comes out and compare that to the fx and you can even shut hyperthreading off and see how much intel is bottlenecked lol.

 

[mosox]

?heck just wait until a real 8 core sandy E comes out and compare that to the fx and you can even shut hyperthreading off and see how much intel is bottlenecked lol.

Sure, can't wait to see a $245.00 8-core Intel CPU. But this won't happen in my lifetime.

 

[grkM3]

Sure, can't wait to see a $245.00 8-core Intel CPU. But this won't happen in my lifetime.

price has nothing to do with it,read what your said.You said intel is bottlenecked becasue AMDs 8 core came close to beating a 4 core.

Is a 2600k bottlenecked also? we are talking about the future here and you seem to think that becasue the amd got gains with an update and still losses to a 4 core chip that intel has an issue.

the bottleneck my friend is the whole BD ARCH.It sucks ballz and is the reason the people left amd.You can have an 8 threaded intel for close to 250 its the 2600k!!!

Its 279 right now at microcenter...so for 35 bucks more you get a chip that will clock to 5+ghz and mop the floor out of a new BD FX chip.

right intel is def the bottleneck lol

 

[carnage10]

Sure, can't wait to see a $245.00 8-core Intel CPU. But this won't happen in my lifetime.

Unless your 90 years old or you plan on dying in the next 10 years, then i think we will definitely see an 8 core Intel CPU under $250 in the next 10 years. Hell, in less than five years we've gone from the first ever quad core costing $1000 to paying less than $180 for Intel's latest offering today. In ~5 years when Octo Cores become the norm, then they will probably cost around the same as what we're paying for Quad Cores today.

 

[greenhawk]

just think how bad Ivy Bridge is going to wipe the floor with them from a performance per watt perspective.

proberly, but as Intel knows it has no competition, just watch the price of Ivy Bridge go up. Even the high asking prices of the AMD chips will look good again on the prive vs performance rankings.

 

[dma0991]

I don't know for sure if Intel would jack up the price for IB but my instincts tell me they would not for the time being. IB is mainly about targeting the mobile space and if Intel were to increase their prices significantly, then the below $1k mark for their ultrabooks is no longer valid.

If IB together with Ultrabook is Intel's answer to ARM then they wouldn't increase the price that much as there is competition from ARM instead of AMD. The fate of desktop counterparts may not be the same but I hope it is the same for all processors in the IB range.

 

[hooflung]

the bottleneck my friend is the whole BD ARCH.It sucks ballz and is the reason the people left amd.You can have an 8 threaded intel for close to 250 its the 2600k!!!

The processor has been out one day officially and you are making that statement. The arch has so much potential you may see Intel doing another copy/paste from AMD in the future. The point is, you don't know what is going on. You aren't an authority and your post has fanboi written all over it.

 

[Concillian]

Yes but with every die shrink you get 3-4% higher IPC too due to various optimizations or at least higher amount of cache.

They will get some IPC from die shrink, but I think they'll use mcuh of the space they've traditionally given to cache to increase the integrated GPU capabilities to catch up with Fusion.

this manufacturing process change may bring about significantly more than the typical benefit from a die shrink though. FinFETS have much more area for conducting, and potential for using multiple in parallel for critical or high current areas. This should significantly reduce heat generation, which should be a very good thing for clock scaling. The interesting part of IB is the potential for a clock speed scaling larger than we've seen in years.

 

[Vesku]

My main point is that by the time people are starting to run things that BD will see nice gains in, then BD v2 should be out. So why not run something that does well NOW price/perf wise and pick up BD v2 IF it turns out to be pretty good on release.

IMO, 8150 should have been a 5 module 10 core chip for ~$250-300. I think I could look past the individual core problems at that ratio.

 

[sm625]

Dick Fuld said he was going to "burn the shorts" weeks before Lehman went bankrupt. A billion was dumped into Solyndra in what was obvious blatant racketeering. While I doubt the same level of racketeering is going on at AMD, the board of AMD still needs to be completely sacked. They should have forced a move to Phenom III well over a year ago, and we should have had 8-16 core K10 chips in our hands by now. Maybe even a 16 core Athlon X16, with no L3 and quad channel memory. What is the point in having L3 if it is slow as molasses? Tell me an athlon X16 wouldnt completely dominate MT benches. And the die size would only be 1.2B transistors.

 

[Nemesis 1]

Ivy Bridge will have an extremely negligible effect on IPC. It is a die shrink.

So says a guy who said BD would have a good IPC improvement . I look for 20% improvement in IB IPC thats about 30% better than BD brought to the table. Whats the smallest improvement intel has made on both ticks and tocks since 06. I believe 10% is the smallest increase. Intel said they did work on the IB cpu portion . and thats what they have told us

 

[Nemesis 1]

I don't know for sure if Intel would jack up the price for IB but my instincts tell me they would not for the time being. IB is mainly about targeting the mobile space and if Intel were to increase their prices significantly, then the below $1k mark for their ultrabooks is no longer valid.

If IB together with Ultrabook is Intel's answer to ARM then they wouldn't increase the price that much as there is competition from ARM instead of AMD. The fate of desktop counterparts may not be the same but I hope it is the same for all processors in the IB range.

Oh this BD thing makes it worse than that . When intel shrank to 32nm not much of a price cut . I fully expected a good pricecut @ 22nm . But with BD as it is . Intel won't lower pricies but margins will rise to 70%