can Intel's Iris PRO inside Broadwell be a low-end discrete video card killer?

[bob345]

Mmmm, not yet. If AMD brings out higher clocked Carrizo SKUs before Zen, it could close that gap pretty easily, especially since Carrizo gets GCN 1.2 with Delta Color Compression, and the Excavator core IPC is higher than Steamroller by a 5-15% margin depending on the workload.

lets hope so. I realy hate seeing amd fall so far behind. I want the real gpu and cpu wars to come back. That would be awesome.

[sgloux3470]

Though a scripted course never covers it all either, and I'd argue it can even cover less in a lot of cases.

 

Depends on if the devs know what they are doing.

 

I'm certainly not trying to say that every ingame benchmark tool is proper, just that doing a human piloted one instead of a perfectly good canned benchmark is ridiculous.

[patrickjp93]

Depends on if the devs know what they are doing.

 

I'm certainly not trying to say that every ingame benchmark tool is proper, just that doing a human piloted one instead of a perfectly good canned benchmark is ridiculous.

It isn't ridiculous, but you have to know exactly what it is you're looking at. That's why I actually prefer JayzTwoCents live benchmarking through gameplay which I can watch.

[patrickjp93]

lets hope so. I realy hate seeing amd fall so far behind. I want the real gpu and cpu wars to come back. That would be awesome.

You should pay very careful attention to how IBM and Nvidia fair against Skylake E5/E7 and Knight's Landing. If they slip, Nvidia will be on the road to collapse, as their low and some of their mid-range sales will be eaten by Iris Pro 7200 and likely the Zen/Zen+ APUs (especially if they come with HBM onboard, at which point Intel will do the same thing). If Nvidia gets knocked out of the HPC space, it will be out of direct competition with Intel (technically they only compete in accelerators since Intel doesn't make dGPUs), leaving it open to a buyout by Intel, something the Nvidia shareholders would love. At that point Intel will have everything it wants and needs to just crush AMD outright and not give a damn who gets the x86 license and AMD's IP. Intel will snatch up Papermaster, Keller, and Koduri, and the competition will be stranded for years enough that Intel can just plod along with microscopic improvements and then make quantum leaps when the competition gets close, permanently keeping the competition crippled.

 

The next 5-6 years will determine the course of semiconductor history, and it all comes down to whether or not IBM and Nvidia can remain competitive. If they can and AMD can stay financially afloat and keep hold of Keller & Koduri, they have a chance to reignite Intel's competitive spirit in the consumer space. If Nvidia and IBM slip out of HPC, it's all over, because ARM isn't going to break in (at least not in time to offset the eradication of PowerPC) now that Avoton and Xeon D are scooping up all the dense microserver and low power server markets that were starting to emerge.

 

If AMD falls apart too early, the other scenario that might reignite the competition is IBM, Nvidia, Apple, or Samsung gets it and retains Papermaster, Keller, and Koduri. In that case, suddenly Intel's dealing with the only engineers that have managed to threaten it backed by a war chest large enough to rival its own. If Nvidia makes the purchase, then Intel will buy up the ATI IP by FTC antitrust rulings, and you'll see Intel take off with dGPUs pretty much immediately, probably a bit shaky at the start. Intel can't afford to crush AMD too early and make it go bankrupt until Nvidia's out of the picture. That said, Nvidia's not pulling punches since it knows AMD's on the financial ropes. The next few years decide if Intel will have competition long-term or not. In the worst case, Samsung, Apple, or IBM take over a bloodied and beaten AMD minus its best engineers with Intel in possession of Nvidia. If that happens Intel will exist as a legitimate monopoly with no one in the industry able to keep pace and secure strong competition. In the best case scenario, Intel gets ATI, Nvidia gets AMD, and competition in the consumer space is reignited.

 

AMD is not going to last as AMD for more than another decade. I just don't see it.

[SageC]

Wow these things kill AMD APUs, one more market AMD has lost.. 

[Notional]

Wow these things kill AMD APUs, one more market AMD has lost.. 

 

How is a 350$ part killing a 150$ market? I don't get it.

[SageC]

How is a 350$ part killing a 150$ market? I don't get it.

 

...You got me there. I guess I didn't really think that through. Here is this better:

 

One more market where AMD has been pushed to the budget option.

[Notional]

...You got me there. I guess I didn't really think that through. Here is this better:

 

One more market where AMD has been pushed to the budget option.

 

Indeed. I assume this will return to status quo, when ZEN apu's hit the market. 14 nm finfet makes room for a lot of apu. But who knows.

[Takata]

Could have included a direct price comparison between the cheap CPU + cheap GPU solution and the Iris Pro solution.

[patrickjp93]

Indeed. I assume this will return to status quo, when ZEN apu's hit the market. 14 nm finfet makes room for a lot of apu. But who knows.

That will depend on whether or not AMD invests in some eDRAM or HBM onboard.  If it doesn't, Intel will pull ahead without a doubt and maintain an unnecessarily wide lead despite having a weaker architecture. AMD is bandwidth-choked, and so is Intel, though Intel definitely doesn't have the same amount of power under the hood.

[Kobz360]

It's heaps good to see Intel putting out some good iGPU's BUT.... guys pls

 

The 5775c and 5675c are like $366 and $276 respectively vs. the 7870K at $137 (Anandtech)..

 

I don't think AMD will be worrying about their budget market being hampered that much. I mean for $360 you can get the 7870K or an i3 + still have more than $200 to spend on a GPU. For $270 you can get the 7870K or even drop it down and get the 860K or something + a 270X or maybe something even slightly better. There's no doubt though that these Iris chips would make a lot of sense for many situations though, I am definitely not disputing that.

[Commander Llama]

It isn't CPUs that are stagnating. It's consumer software design. Intel's flops numbers have multiplied by a factor of 6 in as many years. We have wide vector instructions. Code just has to be allowed to optimize for them. Most consumer software is still compiled as to support all the way back to the minimum of windows XP: the Pentium 3. -O0 vs -Ofast on any C++ compiler is a huge difference. Go take a look at my HPC for Dummies blog here on LTT. I demonstrate very clearly what the differences are.

 

Most CPU stagnation complaints come out of gamers, a large population here, and it's sad but understandable.  Game developers work their asses off, but adding on top of that efficient multi-threading... nope.  It's 2015 and barely any games are multi-threaded.  How long have we had multi-core processors for in the consumer space?  It's so long I don't even remember, and I've had an 8 threaded CPU since 2009 personally.  Thus, gamers bitch about 10% improvements when 40-90% of their CPU isn't being used at all.  The only recent examples that come to mind are BF4 and Cities: Skylines.

[Briggsy]

That will depend on whether or not AMD invests in some eDRAM or HBM onboard.  If it doesn't, Intel will pull ahead without a doubt and maintain an unnecessarily wide lead despite having a weaker architecture. AMD is bandwidth-choked, and so is Intel, though Intel definitely doesn't have the same amount of power under the hood.

 

This ^^ 

 

APU's performance is choked by system memory.

[Notional]

That will depend on whether or not AMD invests in some eDRAM or HBM onboard.  If it doesn't, Intel will pull ahead without a doubt and maintain an unnecessarily wide lead despite having a weaker architecture. AMD is bandwidth-choked, and so is Intel, though Intel definitely doesn't have the same amount of power under the hood.

 

With the investment in HBM, I think AMD will implement it into whatever makes economical sense. I definitely think 2016 will be a very interesting year for AMD's portfolio of products.

[Sakkura]

Most CPU stagnation complaints come out of gamers, a large population here, and it's sad but understandable.  Game developers work their asses off, but adding on top of that efficient multi-threading... nope.  It's 2015 and barely any games are multi-threaded.  How long have we had multi-core processors for in the consumer space?  It's so long I don't even remember, and I've had an 8 threaded CPU since 2009 personally.  Thus, gamers bitch about 10% improvements when 40-90% of their CPU isn't being used at all.  The only recent examples that come to mind are BF4 and Cities: Skylines.

 

It's just economics, dual core CPUs are still being sold, so you hit the biggest market if you focus on playable performance on dual cores.

[zMeul]

With the investment in HBM, I think AMD will implement it into whatever makes economical sense. I definitely think 2016 will be a very interesting year for AMD's portfolio of products.

depending on the operational temps for HBM, it may be impossible to put it on the same die as the APU and air-cool it

might be the reason Fiji is not air-cooled

[Notional]

depending on the operational temps for HBM, it may be impossible to put it on the same die as the APU and air-cool it

might be the reason Fiji is not air-cooled

 

Nah HBM has lower voltage and tdp than GDDR5, so that cannot be it. Fiji is a huge die with 4096 stream processors, so it has about 300w tdp. But it does not need water cooling, as 290x didn't either at 300 w tdp. But water cooling makes it possible for Fiji to be a really small itx size card.

[Brainless906]

there are a lot of pages here are there any comments mentioning that anandtech has the i5-5765c at a little under 300 bucks?

 

As impressive as the performance is i'd hardly call a system with a 300 dollar CPU a "budget build" and the 7870k may be behind performance wise but the i5 is costing you double for that 12 fps

 

I'm sure somebody has pointed this out somewhere in here and been appropriately smashed though..

[zMeul]

Nah HBM has lower voltage and tdp than GDDR5, so that cannot be it. Fiji is a huge die with 4096 stream processors, so it has about 300w tdp. But it does not need water cooling, as 290x didn't either at 300 w tdp. But water cooling makes it possible for Fiji to be a really small itx size card.

that's exactly the problem

Fiji dissipates a lot of heat and the HBM chips are very close to the GPU die itself, making thermal contact via the heatsink - that's a very big problem is HBM operational temps must not reach hi values

[Notional]

that's exactly the problem

Fiji dissipates a lot of heat and the HBM chips are very close to the GPU die itself, making thermal contact via the heatsink - that's a very big problem is HBM operational temps must not reach hi values

 

AMD would never put HBM on an interposer that close, if they could not handle the temps. Remember that Fiji will also come in an air cooled version.

[Briggsy]

AMD would never put HBM on an interposer that close, if they could not handle the temps. Remember that Fiji will also come in an air cooled version.

 

I would guess at separate heatpipes for the HBM chips on an air cooled card, either that or Fiji is going to have a really low temperature throttling point.

 

*edit* for their cpu's/apu's its not an issue because those already have a low throttling point, so HBM on an APU would be just fine for temps.

[zMeul]

Remember that Fiji will also come in an air cooled version.

still to be confirmed; and those 2x8 power connectors make me wonder if ever

[Notional]

still to be confirmed; and those 2x8 power connectors make me wonder if ever

 

Well...

 

[zMeul]

Well...

no, that card is beyond the point; the GTX 780Ti air-cooled cannot use that much power

that card is most likely built for LN2

 

yes:

[Victorious Secret]

Well...

Yea, cherry pick a Kingpin card that is specifically designed (over designed) for additional power for water/LN2 cooling.