Impressive that GTA V Min FPS @ 42FPS A one benefit is that the GPU along side the CPU (on the same chip).

Meanwhile, back on the REAL planet earth, single-threaded, single core (high frequency) advancements are going nowhere with the 100Mhz gains every year. Applications like AutoCAD continue to run at snails pace and I guess 20Ghz would be the minimum to push my models around, i'll be dead before I see either AutoCAD turn multi threaded multi core of a 20Ghz CPU from Intel.

detroit808Meanwhile, back on the REAL planet earth, single-threaded, single core (high frequency) advancements are going nowhere with the 100Mhz gains every year. Applications like AutoCAD continue to run at snails pace and I guess 20Ghz would be the minimum to push my models around, i'll be dead before I see either AutoCAD turn multi threaded multi core of a 20Ghz CPU from Intel.

I feel the same thing about Paradox Interactive games. :(

FrickI feel the same thing about Paradox Interactive games. :(

Cool, I wish it was as easy to find another career than it is find a new types of entertainment ;-)

Compute wise, Iris Pro does something like ~800 FLOPS. While that is only one measure of performance, it's worth noting that one of my 6870s was claimed to do 1TFLOP. While my 6870s aren't exactly new technology, they keep up with games fairly well given their age. Also these CPUs having a 65-watt TDP is pretty awesome too. Given enough time, I suspect Intel will have a very reasonable iGPU solution that completes for mainstream discrete graphics given their R&D budget.

detroit808Meanwhile, back on the REAL planet earth, single-threaded, single core (high frequency) advancements are going nowhere with the 100Mhz gains every year. Applications like AutoCAD continue to run at snails pace and I guess 20Ghz would be the minimum to push my models around, i'll be dead before I see either AutoCAD turn multi threaded multi core of a 20Ghz CPU from Intel.

Give developers a framework for writing highly concurrent 3D applications without taking decades to implement and we'll talk. It's not like there aren't facilities for making applications utilize multiple cores, the problem is figuring out how to solve problems that will benefit from it. The problem is that games are treated as highly serialized workloads and only attempting to use multi-threading for things like physics (which is really just massively parallel ops of the same kind, which is why GPGPU devices excel at doing Physics,) and trying to break down rendering in a way to use multiple cores, but you run into this issue that you need some level of shared state between the concurrent processes which introduces locking and waiting. So while an application might actually be using 73 threads to do its job, there may be so much locking and shared state that it really can't do more than 2 cores worth of computation by virtue of how the threads were tied together.

So if you want to solve the multi-threaded problem for games, then you need to solve the question of how to break a game down into independent pieces that work together in a non-blocking way. So as someone who is a developer and holds a degree in computer science, I'm going to say, "That's easier said than done." There are a lot of considerations for applications that work this way and people don't realize the amount of complexity and issues that are involved.

Also there is a ceiling on clock speeds thanks to the types of transistors we currently use, so it's a question that needs to get answered.

Anandtech already has published their review. Check it here:
www.anandtech.com/show/9320/intel-broadwell-review-i7-5775c-i5-5765c/10

Why do I have a feeling that Moore's Law exists for Intel to know they shouldn't innovate too much opposed to defining our technological limitations and how that reflects on performance of CPU's...

Just think about it, they could be running 10GHz CPU's with insane parallelism that is like 20 times higher than current CPU's, but because Moore's Law is convenient for profit, they just follow it...

I want to see what the mobile Iris Pro will do.

RejZoRWhy do I have a feeling that Moore's Law exists for Intel to know they shouldn't innovate too much opposed to defining our technological limitations and how that reflects on performance of CPU's...

Just think about it, they could be running 10GHz CPU's with insane parallelism that is like 20 times higher than current CPU's, but because Moore's Law is convenient for profit, they just follow it...

Sure, along with a 1000 watt TDP and a liquid nitrogen cooler included, all for $15000! Take off the tin foil hat.

Intel tried to hit that with the pentium 4, and we saw how well that went. hot, slow, and pathetic. To hit those clocks, you need an extremely long pipeline, which is both latent and power hungry. AMD's bulldozer can hit only 5-5.3 GHz without insane cooling or turning off all but 1 core, and even then struggles to keep up with a normal 3.4GHz core i7 cpu in most benchmarks, while pulling in excess of ~220 watts.

There is a REASON we stopped chasing the GHz, it leads nowhere. A core like the one you are proposing would be gigantic, expensive, power hungry, and not all that useful. The insane parallelism would be useless, as a program that takes advantage of parallelism could more easily use an 18 core xeon or a GPU, both of which are much cheaper and are probably more power efficient.

AquinusCompute wise, Iris Pro does something like ~800 FLOPS. While that is only one measure of performance, it's worth noting that one of my 6870s was claimed to do 1TFLOP. While my 6870s aren't exactly new technology, they keep up with games fairly well given their age. Also these CPUs having a 65-watt TDP is pretty awesome too. Given enough time, I suspect Intel will have a very reasonable iGPU solution that completes for mainstream discrete graphics given their R&D budget.

Give developers a framework for writing highly concurrent 3D applications without taking decades to implement and we'll talk. It's not like there aren't facilities for making applications utilize multiple cores, the problem is figuring out how to solve problems that will benefit from it. The problem is that games are treated as highly serialized workloads and only attempting to use multi-threading for things like physics (which is really just massively parallel ops of the same kind, which is why GPGPU devices excel at doing Physics,) and trying to break down rendering in a way to use multiple cores, but you run into this issue that you need some level of shared state between the concurrent processes which introduces locking and waiting. So while an application might actually be using 73 threads to do its job, there may be so much locking and shared state that it really can't do more than 2 cores worth of computation by virtue of how the threads were tied together.

So if you want to solve the multi-threaded problem for games, then you need to solve the question of how to break a game down into independent pieces that work together in a non-blocking way. So as someone who is a developer and holds a degree in computer science, I'm going to say, "That's easier said than done." There are a lot of considerations for applications that work this way and people don't realize the amount of complexity and issues that are involved.

Also there is a ceiling on clock speeds thanks to the types of transistors we currently use, so it's a question that needs to get answered.

The TSX extensions was a attempt at giving better parallelism, if they had added more atomic actions content sync over threads would be a bit easier at least. but Amdahls law still applies.

RejZoRWhy do I have a feeling that Moore's Law exists for Intel to know they shouldn't innovate too much opposed to defining our technological limitations and how that reflects on performance of CPU's...

Just think about it, they could be running 10GHz CPU's with insane parallelism that is like 20 times higher than current CPU's, but because Moore's Law is convenient for profit, they just follow it...

Well, they are following mores law, they are just pouring the transistor budget into the GPU, and not the CPU, also, 128 MiB L4, not that is helps much.

TheinsanegamerNSure, along with a 1000 watt TDP and a liquid nitrogen cooler included, all for $15000! Take off the tin foil hat.

Intel tried to hit that with the pentium 4, and we saw how well that went. hot, slow, and pathetic. To hit those clocks, you need an extremely long pipeline, which is both latent and power hungry. AMD's bulldozer can hit only 5-5.3 GHz without insane cooling or turning off all but 1 core, and even then struggles to keep up with a normal 3.4GHz core i7 cpu in most benchmarks, while pulling in excess of ~220 watts.

There is a REASON we stopped chasing the GHz, it leads nowhere. A core like the one you are proposing would be gigantic, expensive, power hungry, and not all that useful. The insane parallelism would be useless, as a program that takes advantage of parallelism could more easily use an 18 core xeon or a GPU, both of which are much cheaper and are probably more power efficient.

Lol you say that if CPU's/Processors are going to be stuck on silicon for eternity o_O

Intel Leads a New Era of Computing Experiences

not if u have an i7 2600K. Gotta go last decade.

Interesting to see what the faster cores do for the integrated graphics, AMD needs to see this as a shot across the bow if you will, as this chip could easily replace the AMD chips in the consoles and perform way better.

TheinsanegamerNSure, along with a 1000 watt TDP and a liquid nitrogen cooler included, all for $15000! Take off the tin foil hat.

Intel tried to hit that with the pentium 4, and we saw how well that went. hot, slow, and pathetic. To hit those clocks, you need an extremely long pipeline, which is both latent and power hungry. AMD's bulldozer can hit only 5-5.3 GHz without insane cooling or turning off all but 1 core, and even then struggles to keep up with a normal 3.4GHz core i7 cpu in most benchmarks, while pulling in excess of ~220 watts.

There is a REASON we stopped chasing the GHz, it leads nowhere. A core like the one you are proposing would be gigantic, expensive, power hungry, and not all that useful. The insane parallelism would be useless, as a program that takes advantage of parallelism could more easily use an 18 core xeon or a GPU, both of which are much cheaper and are probably more power efficient.

And all of the sudden Core architecture appeared out of nowhere. I think you should take off that tin foil hat...

RejZoRAnd all of the sudden Core architecture appeared out of nowhere. I think you should take off that tin foil hat...

What? It didn't appear out of nowhere. It was an enhanced version of core M, which was the continuation of the p6 core. and pentium 4 was much slower than the p6 core in the first place, that core 2 looked astonishingly good. core 2 was the result of years of incremental work that was brought to fruition, not something that appeared overnight.

If you look at the first generation core i series from 2007, and the new haswell chips today, there is a remarked increase in IPC, roughly ~5% per generation. moving from nehalem to haswell brings a remarked improvement in performance per clock, not to mention the power savings, which is where most of the work is going. the performance/watt of modern chips is much better than first gen i series chips.

Intel accidentally struck gold with nehalem, and again with sandy bridge. Improving from that is going to be very difficult with currently available technology. skylake is supossed to bring about big IPC improvements, but I'll believe it when I see it. Silicon is starting to reach the end of it's usefullness on that front.

Lets face it and see critically on the problem.

As of today, we have more computerpower than ever before, not only as compays or as just normal people.

I just ask do we need more of that ?, I dont think so since development if going to smaller portiable and handheld devices.
As a consumer, we have enough, more than enough and theres still the need to get more and more, its insane. Even avarage Joe who has a 3 year computer still have more than enough CPU power for everyday need and he is happy. His machine can do what he wants and when he wants it, so where is his need for upgrading ?, unless it brakes down.

For us who likes to have new toys to play with well thats a different matter, but the trouth is, that CPU power is enough, its all about grafics when you play games.

Some people set consuming power high on their list and some dosent, freedom of choice. If AMD's new Zen CPU which comes in 2016 is just as good or better and uses less power than my I7-5820K that I have now, I will change in a heartbeat.

EboLets face it and see critically on the problem.

As of today, we have more computerpower than ever before, not only as compays or as just normal people.

I just ask do we need more of that ?, I dont think so since development if going to smaller portiable and handheld devices.
As a consumer, we have enough, more than enough and theres still the need to get more and more, its insane. Even avarage Joe who has a 3 year computer still have more than enough CPU power for everyday need and he is happy. His machine can do what he wants and when he wants it, so where is his need for upgrading ?, unless it brakes down.

For us who likes to have new toys to play with well thats a different matter, but the trouth is, that CPU power is enough, its all about grafics when you play games.

Some people set consuming power high on their list and some dosent, freedom of choice. If AMD's new Zen CPU which comes in 2016 is just as good or better and uses less power than my I7-5820K that I have now, I will change in a heartbeat.

I would agree with you in 99% of the cases, but i can tell you that some games realy likes cpu power, i can compare Civ 5 on my 5820k at 4.3 Ghz, with 2 290x in CF to my i7 920 at 2.6 gHz with 2 hd 6950 in cf and the game looks as good on both machines, but the turn time is noticeably lower on the 5820k.

Then again, this is a fringe case, but it is always nice to have more cpu power.

It should be like that:

Intel Stalls New Era of Computing Experiences

8 cores and only 1 working... progress a la Intel. It should be CPU job to utilize whole potential of CPU not just software.

BrusfantometI would agree with you in 99% of the cases, but i can tell you that some games realy likes cpu power, i can compare Civ 5 on my 5820k at 4.3 Ghz, with 2 290x in CF to my i7 920 at 2.6 gHz with 2 hd 6950 in cf and the game looks as good on both machines, but the turn time is noticeably lower on the 5820k.

Then again, this is a fringe case, but it is always nice to have more cpu power.

Underclock that 5820 to 2.6 or bump 920 to stock 5820 (which can be done easily), and what? No difference. I have X58 and I'm only limited by Gen 2.0 PCI-Ex for storage, because of that switch is imminent. to 5930 (40 lanes). Other than this there is no reason to upgrade. And I predict that it will be last upgrade ever done. For only gaming X99 is pointless. VGAs cannot saturate x8 Gen 2 let alone Gen 3. Storage on the other hand with new RAID controllers can and it forces me to move (SSD arrays, slowly but surely).

ypsylonUnderclock that 5820 to 2.6 or bump 920 to stock 5820 (which can be done easily), and what? No difference. I have X58 and I'm only limited by Gen 2.0 PCI-Ex for storage, because of that switch is imminent. to 5930 (40 lanes). Other than this there is no reason to upgrade. And I predict that it will be last upgrade ever done. For only gaming X99 is pointless. VGAs cannot saturate x8 Gen 2 let alone Gen 3. Storage on the other hand with new RAID controllers can and it forces me to move (SSD arrays, slowly but surely).

Point was not about the IPC of the different cores, but more that the game benefits from the extra cpu power, not defining if it comes from the newer core or higher clock.
My choice for X99 was enough lanes for at least two cards and 4 lanes for a m.2 SSD

ypsylonStorage on the other hand with new RAID controllers can and it forces me to move (SSD arrays, slowly but surely).

Software RAID is not depended on the hardware and is cheaper or free.

Brusfantometand 4 lanes for a m.2 SSD

Why does everyone want to have SSD in M.2 slot, I am satisfied with SSD connected in SATA port and if in the future we could have an option to choose NVMe instead of AHCI (I know NVMe is used for PCIe but why shouldn't a version of NVMe also be used for SATA SSDs) for every SATA port it would be perfect. PCIe is faster but sequential speeds are not everything and when SSD can do 32Gb/s for random writes I will be very happy.
You just cannot have so many different ports and standards and I think SATA is not expandable and you cannot put HDD in M.2 slot.

I wanted it to get rid of S-ata cables. they fall out form the SSD over time, the M.2 reduces cable clutter.

BrusfantometI wanted it to get rid of S-ata cables. they fall out form the SSD over time, the M.2 reduces cable clutter.

I can understand your concern, but SATA drives are more versatile, it is easier to build NAS with SATA cables than M.2 because there are not many M.2 slots on motherboard and SATA is more universal than M.2. If we all switch to SSDs, I would not mind seeing only M.2 slots (probably off motherboard) but that will not happen in the coming 10 years.

ypsylon8 cores and only 1 working... progress a la Intel. It should be CPU job to utilize whole potential of CPU not just software.

I'm leveraging my experience as a software dev and as a person with a degree in Comp Sci on this matter because it's highly applicable and the record needs to be set straight before you start putting weird ideas in people's heads and I'll explain why.

This is easier said than done. You can't simply make all workloads parallel because most code written is serial, there are dependencies between instructions and as a result, would need to share state and memory. The issues here is that overhead of making it parallel might introduce less performance if there is enough coordination that has to occur because most applications will do work to build up data, not build it in parallel, as a result, instructions will depend on the changes from the last (see serializability). It is the developer's job to write code in a way that can leverage hardware when it's available and to know when to apply it because not everything can be done on multiple cores by virtue of the task being done. The state problem is huge and locking will destroy multi-threaded performance. As a result, many successful multi-threaded systems have parts of an application decoupled with queues placed inbetween tasks that much occur serially. This all sounds find and dandy, but now you're just passing state through essentially a series of queues and any work will have to wait if any part of the pipeline slows down (so you're still limited by the least parallel part of your application.) So while you inherently get parallelism by doing this, you add the side effect of increasing latency and not knowing the state of something at any level other than the one it's operating at. So in the case of games, they need to react quickly to less latency means either lower frame rate or input lag.

So while your comment makes sense from looking at it from a high level, it makes absolutely no sense at a low level because that isn't how computers work under the hood and you can't simply break apart an application and make it multi-threaded. It simply doesn't work that way.

I would argue that developers need better tools for easily implementing games that can utilize multiple cores but, languages (like Java, Clojure, C#, etc.) have great constructs for multi-threaded programming. The question isn't if you can do it, the question what is the best way to do it and no one really knows.

I tell people this all the time: "If it were that easy, it would have been done already!"

Lastly, this is already done in processors to some extent at the opcode/microcode level. It's called pipe-lining and most CPUs (which are super scalar,) work this way. Has for a long time, but this is done at the instructional level in an individual core because we're still talking about serial workloads (which is what you're generic application usually is.)

BrusfantometI wanted it to get rid of S-ata cables. they fall out form the SSD over time, the M.2 reduces cable clutter.

All 8 of the SATA ports in my tower are used. One M.2 and be done with it, but it's really not designed for mass storage. SATA will always have a place in my book until something better rolls around that doesn't require the device to be attached to the motherboard. (Imagine mounting a spinning drive to a motherboard, that makes for some pretty funny images. :p

Side note: I had more issues with IDE cables than SATA, so I'm not complaining.

AquinusI'm leveraging my experience as a software dev and as a person with a degree in Comp Sci on this matter because it's highly applicable and the record needs to be set straight before you start putting weird ideas in people's heads and I'll explain why.

This is easier said than done. You can't simply make all workloads parallel because most code written is serial, there are dependencies between instructions and as a result, would need to share state and memory. The issues here is that overhead of making it parallel might introduce less performance if there is enough coordination that has to occur because most applications will do work to build up data, not build it in parallel, as a result, instructions will depend on the changes from the last (see serializability). It is the developer's job to write code in a way that can leverage hardware when it's available and to know when to apply it because not everything can be done on multiple cores by virtue of the task being done. The state problem is huge and locking will destroy multi-threaded performance. As a result, many successful multi-threaded systems have parts of an application decoupled with queues placed inbetween tasks that much occur serially. This all sounds find and dandy, but now you're just passing state through essentially a series of queues and any work will have to wait if any part of the pipeline slows down (so you're still limited by the least parallel part of your application.) So while you inherently get parallelism by doing this, you add the side effect of increasing latency and not knowing the state of something at any level other than the one it's operating at. So in the case of games, they need to react quickly to less latency means either lower frame rate or input lag.

So while your comment makes sense from looking at it from a high level, it makes absolutely no sense at a low level because that isn't how computers work under the hood and you can't simply break apart an application and make it multi-threaded. It simply doesn't work that way.

I would argue that developers need better tools for easily implementing games that can utilize multiple cores but, languages (like Java, Clojure, C#, etc.) have great constructs for multi-threaded programming. The question isn't if you can do it, the question what is the best way to do it and no one really knows.

I tell people this all the time: "If it were that easy, it would have been done already!"


All 8 of the SATA ports in my tower are used. One M.2 and be done with it, but it's really not designed for mass storage. SATA will always have a place in my book until something better rolls around that doesn't require the device to be attached to the motherboard. (Imagine mounting a spinning drive to a motherboard, that makes for some pretty funny images. :p

Side note: I had more issues with IDE cables than SATA, so I'm not complaining.

Having done some programming I agree entirely. For multithreaded workloads with dependencies you have to have a parent thread start and dispatch child processes or threads to perform the work while the parent thread synchronizes the results, and that cannot be done on hardware as the hardware has no idea of the actual work being done, it only sees a string of binary. So at the point where we have a parent thread and at least one child thread, and so if we break it down to human language.


Parent thread launches on core 0, it copies in a large dataset for comparison, it then launches a child thread that starts at the top of the thread, and it starts at the bottom, and it then has to check up on a flag being set by the child thread to see if it found a match thus reducing its performance. Try to find customer 123456789 in a dataset that is alpha numeric and customers numbers are randomized so either you have to sort, search the whole table, or you could have it check using a smarter algorithm. Depending on the size of the dataset and processor speed it might be as fast to sort on a single core as it is to search and compare on multiple cores.