• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Intel "Haswell" Speculation thread

Page 17 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Wake me up when you have more than "in the future"...

In the future I still don't wan't to waste CPU-diespace on a IGP.
I'll buy a GPU...get over that I don't think IGP's are of any use for me.

Hence my money will only buy CPU's that is free of crappy IGP's

Hey, that's completely your prerogative. You can spend your money any way you want. I'm just giving a heads up straight up 2 core, 4 core, X core CPUs isn't where tech is going. Just look at the supercomputers, how did that happen right?
 
Hey, that's completely your prerogative. You can spend your money any way you want. I'm just giving a heads up straight up 2 core, 4 core, X core CPUs isn't where tech is going. Just look at the supercomputers, how did that happen right?

Too bad Haswell-E says otherwise too you.

I'd rather have 2-4 cores more and no IGP..and I'll pay for such a solution.
 
Too bad Haswell-E says otherwise too you.

I'd rather have 2-4 cores more and no IGP..and I'll pay for such a solution.

Well, its not just a plain GPU die area to CPU cores conversion.

Without more drastic changes, on IB for example you could get maybe ½ an extra core from the iGPU before you would have to penalize the current cores or increase TDP and the cooling solution.

The iGPU is actually a bonus in terms of cooling.
 
Well, its not just a plain GPU die area to CPU cores conversion.

Without more drastic changes, on IB for example you could get maybe ½ an extra core from the iGPU before you would have to penalize the current cores or increase TDP and the cooling solution.

The iGPU is actually a bonus in terms of cooling.

Not interested in TDP, perf/watt or any other FOTM metric.
In a CPU...I care about CPU performance.
If I need video power...I'll add a GPU...that will run circles (performance wise) around any crappy IGP.

I have no desire or use for onboard crappy stuff.
I don't use onboard sound or onboard LAN either, I diable those and put in a NIC and a soundcard...and I have even less use for the wanna-bee-GPU's...
 
99% of users will be fine with an APU, the real boys(1%) will always need more CPU performance/threads and discrete GPUs 😉

No matter how fast the APUs will become(fused, more apps to take advantage off etc), a CPU + Discrete GPU will always be faster in both apps and Games.
 
99% of users will be fine with an APU, the real boys(1%) will always need more CPU performance/threads and discrete GPUs 😉

No matter how fast the APUs will become(fused, more apps to take advantage off etc), a CPU + Discrete GPU will always be faster in both apps and Games.

QFT...and thus I will alwys run a CPU + GPU...not an CPU with an IGP...
 
Not interested in TDP, perf/watt or any other FOTM metric.
In a CPU...I care about CPU performance.
If I need video power...I'll add a GPU...that will run circles (performance wise) around any crappy IGP.

I have no desire or use for onboard crappy stuff.
I don't use onboard sound or onboard LAN either, I diable those and put in a NIC and a soundcard...and I have even less use for the wanna-bee-GPU's...

I will second this as I also do not like the on board IGP.

In fact, I have built a couple of Intel Xeon E3-1240 V2 machines since I could not get rid of the on board IGP. The Xeon E3-1240 V2s work fine with z77 chip sets even though support for them is not officially listed.
 
I have no desire or use for onboard crappy stuff.
I don't use onboard sound or onboard LAN either, I diable those and put in a NIC and a soundcard...and I have even less use for the wanna-bee-GPU's...

When you wake up from 2003, let us know.
 
No matter how fast the APUs will become(fused, more apps to take advantage off etc), a CPU + Discrete GPU will always be faster in both apps and Games

So what you are trying to say is, that if the IGP was able to offload some of the FP operations that would otherwise have to go through the CPU, and in turn speed up the entire application, while leaving the dedicate GPU alone for just graphic processing, you would not wlecome that?

I am starting to look at the IGP as more of a co-processor of sorts. It may be able to perform some single percision FP operations faster than an Intel CPU core. And it comes free with new CPUs. I will take it.
 
I am starting to look at the IGP as more of a co-processor of sorts. It may be able to perform some single percision FP operations faster than an Intel CPU core. And it comes free with new CPUs. I will take it.

- I would like a perspective on how avx2+ will complement the igpu on haswell and onwards .. what kind of synergies are we looking at.
Taking it a notch firuther, what kind of synergies will *that* have with a powerfull descrete card?
 
Hence my money will only buy CPU's that is free of crappy IGP's

You don't understand economics of these chip companies do you? The fact that having an iGPU allows more than 2/3 of users to be fine means its essentially paid off. You get NO advantage by having a seperate, iGPU-less part.

If Intel and/or AMD didn't have iGPUs, the ASPs(Average Selling Prices) of the chip would have to go down, usually in proportion to how much cheap discrete cards cost. That means it's well worth it for them and YOU to continue iGPU development.
 
You don't understand economics of these chip companies do you? The fact that having an iGPU allows more than 2/3 of users to be fine means its essentially paid off. You get NO advantage by having a seperate, iGPU-less part.

If Intel and/or AMD didn't have iGPUs, the ASPs(Average Selling Prices) of the chip would have to go down, usually in proportion to how much cheap discrete cards cost. That means it's well worth it for them and YOU to continue iGPU development.

You don't get it do you?
I dislike the idea of CPU+IGP.
So I will not buy any products that fit this description.

Live with it.
 
All LGA 1155 CPUs have an IGP. The IGP is disabled on some (such as Xeons), but it is still physically there.

As for you not liking the idea of CPU+IGP - you won't be upgrading much beyond SB-E. The IGPs Intel uses are on the CPU die now. Intel has no reason to modify the die and release it without the IGP.
 
So what you are trying to say is, that if the IGP was able to offload some of the FP operations that would otherwise have to go through the CPU, and in turn speed up the entire application, while leaving the dedicate GPU alone for just graphic processing, you would not wlecome that?

Why they would have to go to the CPU, you forgetting that i will have a monster(vs the iGPU) discrete GPU that it would be able to do the job of the iGPU much faster saving time and energy for me. No i dont need the iGPU, my discrete GPU will always have more performance than the iGPU.

I am starting to look at the IGP as more of a co-processor of sorts. It may be able to perform some single percision FP operations faster than an Intel CPU core. And it comes free with new CPUs. I will take it.

We already have an FP execution unit (co-processor) in the CPU, i dont need it to be larger because i already use an even larger co-processor, my discrete GPU.

As i have said before, for 99% of users it will be a nice thing to have. But for the big boys its a waste of resources.
 
You don't understand economics of these chip companies do you? The fact that having an iGPU allows more than 2/3 of users to be fine means its essentially paid off. You get NO advantage by having a seperate, iGPU-less part.

I will reverse that and say that we could have 6-8 core Desktop CPUs today at the same prices of current Socket 1155 CPUs if there wasn't any Intel APU in the market.

If Intel and/or AMD didn't have iGPUs, the ASPs(Average Selling Prices) of the chip would have to go down, usually in proportion to how much cheap discrete cards cost. That means it's well worth it for them and YOU to continue iGPU development.

That is not true, see above. You would have a high ASP simple because you would have more cores, cache or both in the same die space as an Intel CPU + iGPU today.

The CPU + iGPU is a strategic move Intel took simple to counter AMDs APUs project. They cannot stay outside of this for the low end and middle market segments.

High End Desktop/workstations and Server will continue to have iGPU-less CPUs in the future, they maybe will also have APUs as well but never one or the other alone.
 
All LGA 1155 CPUs have an IGP. The IGP is disabled on some (such as Xeons), but it is still physically there.

As for you not liking the idea of CPU+IGP - you won't be upgrading much beyond SB-E. The IGPs Intel uses are on the CPU die now. Intel has no reason to modify the die and release it without the IGP.

Xeons got IGP on. Its called P4000.
 
Wake me up when you have more than "in the future"...

In the future I still don't wan't to waste CPU-diespace on a IGP.
I'll buy a GPU...get over that I don't think IGP's are of any use for me.

Hence my money will only buy CPU's that is free of crappy IGP's

I agree with your sentiment. I'd rather see that die space put to use on another core (or two) or perhaps more cache.

But, I think you and I will find very few purchasing options going forward. Intel will continue to integrate more and more into their dies till they have a full SoC.
 
SoC? What's that?

System on a chip. It's what all the smartphone processor groups are already doing. Basically the ability to integrate whatever willy nilly stuff you want on the same die. This is beyond just 2 core vs 4 core. This is putting in any GPU, any mix of CPUs, any interface, any radio, any type of memory, any display hardware. Basically you get to mix and match whatever crap you want (and ideally to remove it later when it's no longer relevant).

It's not quite the whole "fusion" idea where everything is blended into one big blob but it mostly gets the job done and plus you get to have several different groups develop different hardware and you just mix and match the good ones together to make a product.
 
We already have an FP execution unit (co-processor) in the CPU, i dont need it to be larger because i already use an even larger co-processor, my discrete GPU.

Which runs what instruction set exactly? And is good at running all FP code?
 
Then most likely in about two or three years there will not be a x86 CPU for you to buy.

IF and if...I purchase in the present...not the future...you keep promoting the midget-wanna-be-GPU...then I will run a real GPU...and laugh 🙂
 
Back
Top