Core i7-4770K is performance crippled

[TuxDave]

Yeah. I don't know any names but I hear that some business customers need performance above and beyond all reasonable cost and will overclock/overvolt the crud out of their processors and replace them every month.

 

[Lepton87]

Yeah. I don't know any names but I hear that some business customers need performance above and beyond all reasonable cost and will overclock/overvolt the crud out of their processors and replace them every month.

They were even some 4.4GHz stock westmere Xeons, some companies in some micro-transactions business require the fastest single threaded performance they can get, I guess those overclock. If someone knows something more about this please expand. I have a black hole in my memory.

 

[jpiniero]

They were even some 4.4GHz stock westmere Xeons, some companies in some micro-transactions business require the fastest single threaded performance they can get, I guess those overclock. If someone knows something more about this please expand. I have a black hole in my memory.

Those were golden samples though (and only dual core on what was a Quad Core processor), and not really in any volume. Intel doesn't even acknowledge it on Ark.

 

[StrangerGuy]

You clearly do not understand exactly what TSX brings to the table then. For starters it will make it easier to develop games with more threads, and hence able to utilize more cores. If you want more games to start using 8/10/12 threads, TSX can make it easier and cheaper to develop.

99% of existing PCs doesn't have TSX or AVX2, tons of those users will never buy a x86 PC again and Intel still thinks it can squeeze water out of a dried nut by persuading people to spend extra for stuff that requires mass adoption to work.

"Myopic" would be an understatement. Then again, this is the same company that missed the mobile boat completely for the past 6 years.

 

[Idontcare]

99% of existing PCs doesn't have TSX or AVX2, tons of those users will never buy a x86 PC again and Intel still thinks it can squeeze water out of a dried nut by persuading people to spend extra for stuff that requires mass adoption to work.

"Myopic" would be an understatement. Then again, this is the same company that missed the mobile boat completely for the past 6 years.

I agree.

Intel is doing the "tortoise and the hare" thing again where after having raced well ahead of the competition they are now busying themselves entertaining all manner of internal distractions concocted by the marketing dept and it is going to open the door for the hare to pass them by once again if they aren't careful.

Having it come to a head simultaneous with handing over the reigns to a new CEO is even more risky as focus is surely to be lost for a couple quarters there as well.

Its not all doom and gloom, but it is certainly all the yellow flags of impending doom and gloom type stuff so I'm not going to be surprised at all if 14nm comes around and it is all the more stuffed because the yellow flags are right here in front of us already.

 

[R0H1T]

I agree.

Intel is doing the "tortoise and the hare" thing again where after having raced well ahead of the competition they are now busying themselves entertaining all manner of internal distractions concocted by the marketing dept and it is going to open the door for the hare to pass them by once again if they aren't careful.

Having it come to a head simultaneous with handing over the reigns to a new CEO is even more risky as focus is surely to be lost for a couple quarters there as well.

Its not all doom and gloom, but it is certainly all the yellow flags of impending doom and gloom type stuff so I'm not going to be surprised at all if 14nm comes around and it is all the more stuffed because the yellow flags are right here in front of us already.

Wait a sec, ain't Intel supposed to be the hare in this case

 

[MisterMac]

I agree.

Intel is doing the "tortoise and the hare" thing again where after having raced well ahead of the competition they are now busying themselves entertaining all manner of internal distractions concocted by the marketing dept and it is going to open the door for the hare to pass them by once again if they aren't careful.

Having it come to a head simultaneous with handing over the reigns to a new CEO is even more risky as focus is surely to be lost for a couple quarters there as well.

Its not all doom and gloom, but it is certainly all the yellow flags of impending doom and gloom type stuff so I'm not going to be surprised at all if 14nm comes around and it is all the more stuffed because the yellow flags are right here in front of us already.

...in mobile?

...in desktop?

... in server\workstation?


In our little enthusiast\mainstream segment - they're clearly playing with fire and opening the door for AMD to get closer.
Clearly they're either not worried they can pedal to the metal if AMD gets close enough - or they've internally just litterally given up and will now focus on perf\watt.

(Kind of like AMD is "trying" to do, ya know?)

The rest? really?

you think AMD has the option to produce 12 core, 14 core full fledged huge server cores?
and beat raw performance and price to boot?


or do LTE SoC's that MIGHT end up in phones?

Intel has alot to prove in the phone segment - but they have some direct working marketable silicon in action atleast.
Unlike AMD.

Don't get me wrong i hate it - but AMD somehow taking over the server market or getting far enough with GF to enter the really low power markets seems more far fetched than them getting close to ST perf of Big Core intel.

 

[mrmt]

Intel is doing the "tortoise and the hare" thing again where after having raced well ahead of the competition they are now busying themselves entertaining all manner of internal distractions concocted by the marketing dept and it is going to open the door for the hare to pass them by once again if they aren't careful.

What's the point in squeezing more performance of your architecture when your chip will be out of the most popular form factors of the decade?

 

[guskline]

IDC: Interesting post. I only know anything about the Haswell chip from what I have read from reviews. I have no personal experience. The number of "doom and gloom" posts is interesting. My qustion is "what did you expect?". Unlike Bulldozer that was under wraps and all we had was marketing BS and that silly midnight Guiness record in some lab, the Haswell has been written about and the actual cpu disclosure is no surprise. No huge improvement, just incremental. Runs hotter than SB (so does IB) and has more extentions.

The tortiose and hare comment may have some validity. I think what IDC commented on recently about buying a Kindle and moving on may be more worrisome to the like of Intel.

Without being overdramatic, AMD has been fighting for it's life the last few years with little financial reserve. Intel, on the other hand has done well financially. HOWEVER, they both face the "Elephant in the room" which is the incredible switch to mobile products - phones, tablets etc. My uneducated gut tells me that Intel is switching its base to be a part of this but in a long term way. AMD is also trying to be a player but their financial house is so bad that they have a much more difficult task ahead to be relevant as a cpu manufacturer. They will be a chip designer, not manufacturer.

Haswell gives Intel a platform to extend the life of their processor to the business world and stave off the onslaught of tablets.

For us enthusiast "nuts"? Haswell really is not a big deal. Also, for the gamers on this forum, unless you are using a 2011 socket & cpu the price of mb & cpu has not increased dramatically. The GPUs have. I think Intel may be saying to us "look if you are going to spend a grand on a Titan or 7990" or even $500- $700 on your GPU, step up to a socket 2011 and play in the big leagues. AMD's response can only be the Piledriver because their real focus was the console market to keep the money pipeline alive.

 

[beginner99]

The TSX spec clearly states that any code that utilizes these new instruction will also work on CPUs without TSX enabled.....it will just not benefit from it hence not run as fast. So there is no reason why developers can not start using it today. It will work on ANY CPU, but only faster on TSX enabled CPUs.

Of course they can start working on it and I know about HLE and fallback. It makes sense in server world but not for consumers and hence the 4770k is fine without it.

if you quote with whole context you would see that my comment was about games which are mostly GPU limited anyway but more importantly why should a game developer invest even a single second for a feature that can be used by like 0.01% of his customers?

 

[Idontcare]

Wait a sec, ain't Intel supposed to be the hare in this case

Derp! You are right, my bad

you think AMD has the option to produce 12 core, 14 core full fledged huge server cores?
and beat raw performance and price to boot?


or do LTE SoC's that MIGHT end up in phones?

Intel has alot to prove in the phone segment - but they have some direct working marketable silicon in action atleast.
Unlike AMD.

Don't get me wrong i hate it - but AMD somehow taking over the server market or getting far enough with GF to enter the really low power markets seems more far fetched than them getting close to ST perf of Big Core intel.

There once was a time when Intel was the Calxeda of the microprocessor market.

At the time there were companies who had already been dominating the space for decades already, who had nice fat margins and a performance lead that was second to no other.

And then Intel came along, as scrappy as it was at the time, and completely took their lunch money and put nearly every single one of them out of business.

Now here we are, 20-30 yrs later, a whole new generation of lethargic and lazy management is in control (hell they were just kids when Intel's management was laying the groundwork for becoming an 800lb gorilla)...and today's management are doing exactly the same thing that DEC's management did to set the foundation for yet another cycle of smaller scrappier companies to come along and eat their lunch.

I'm not at all worried about who exactly is going to be "the next Intel", that is for wallstreet speculators to fret over predicting, but I am absolutely comfortable with placing my bets on history repeating itself and Intel's complacency and managerial hierarchy being their own worst enemy.

What's the point in squeezing more performance of your architecture when your chip will be out of the most popular form factors of the decade?

Said a Fairchild semiconductor exec circa 1974....and a Cray exec circa 1984...and a DEC corporate exec circa 1994...and a SUN exec some 10yrs later...and an Intel exec 10yrs later still.

Go back in recorded history as far as you like and one trait is shared 100% across all business entities regardless of nationality and continent - eventually complacency and hubris combine forces and the outcome is the same over and over again.

Intel is clearly no longer paranoid in the desktop and server markets, and only the paranoid survive.

 

[Durp]

My qustion is "what did you expect?". Unlike Bulldozer that was under wraps and all we had was marketing BS and that silly midnight Guiness record in some lab, the Haswell has been written about and the actual cpu disclosure is no surprise.

I know this was directed at IDC but personally I expected the slight bump in IPC that we got but with solder instead of the TIM and glue. The reason I expected this is because an Intel employee hyped up Haswell.

His words:
" If you like to overclock, Haswell is worth it (can't tell you why but read the Haswell Anandtech preview very carefully for buried treasure)."

Where was the treasure?

Not even close to the JF-AMD/bulldozer nightmare but there was some fake hype here too.

 

[mrmt]

Go back in recorded history as far as you like and one trait is shared 100% across all business entities regardless of nationality and continent - eventually complacency and hubris combine forces and the outcome is the same over and over again.

Intel is clearly no longer paranoid in the desktop and server markets, and only the paranoid survive.

Intel is being paranoid, more paranoid than ever I believe. They are trying to not be a second IBM, that gave away the multi-billion PC market to save the bacon of their single-billion Mainframe market. They are throwing everying + mother on mobile. I don't think they are being complacent, I think they are phasing out of the desktop market, they are so paranoid about mobile that they are forgeting everything else.

What we are seeing here is a glimpse of what Mark Borh said in a interview some time ago:

http://semimd.com/blog/2012/10/18/deep-inside-intel/

Bohr= Intel is very serious about getting into the smart phone and tablet markets. We are a very different company from what we were five or six years ago. We are developing process technologies, but also products, that span a much wider range of performance and power than anywhere in our history. We’re not just after the high-performance desktop. We’re developing products that support 100-watt server chips down to sub-1 watt smart phone chips.

(...)

SMD= How many cores will be required in the future?
Bohr= It depends on the market. In the server market, the more cores you can pack on the better. But in desktops, laptops and smart phones, there’s probably a limit to how many cores are practical. It’s not one. It’s probably several.

SMD= But less than eight?
Bohr= Yes, probably less than eight. But when you talk about the number of cores and computing engines, it depends on whether you’re dealing with traditional computing tasks where four cores are better than two cores. If you’re talking about execution engines in a graphics processor, clearly you want more cores.

(...)


SMD= Has the priority for what you’re designing into a chip changed? Is it still all about performance, or has power overtaken that?
Bohr= Ten or 15 years ago, performance was the main goal in developing a new process technology. That really has gone away as the No. 1 priority. We still strive to provide a performance boost with each new technology, but there’s much more emphasis on improving power or efficiency on each new generation. We do that by reducing active power for the work a chip does. That’s a much more important goal for us today. Part of the reason is that the market has shifted from desktop applications to more mobile products. The first transition was from desktops to laptops. Now the move is to put things into smart phones. Today’s consumer wants computing power he can hold in his hand in the form factor of a smart phone and a tiny battery. He wants the performance he had on his laptop only three or four years ago. That’s what we shoot for.

Btw, that interview is full of hidden nuggets. It's been a year, but it's a worthy reading.

 

[blackened23]

His words:
" If you like to overclock, Haswell is worth it (can't tell you why but read the Haswell Anandtech preview very carefully for buried treasure)."

Actually, If I remember correctly those weren't his words at all. He characterized overclocked Haswell to be "interesting" due to the new bclk overclocking feature, but didn't really go further than that. Of course everyone interpreted that to mean Haswell being what Sandy Bridge was a few years back.

That said, people obviously expected a little bit more with Haswell overclocking - The desktop user in me wanted more, I don't disagree there at all. As things are, though, it's still an appreciable IPC increase over IVB which will offset any potentially lower OC anyway compared to SB or IVB.

 

[ShintaiDK]

Intel is being paranoid, more paranoid than ever I believe. They are trying to not be a second IBM, that gave away the multi-billion PC market to save the bacon of their single-billion Mainframe market. They are throwing everying + mother on mobile. I don't think they are being complacent, I think they are phasing out of the desktop market, they are so paranoid about mobile that they are forgeting everything else.

What we are seeing here is a glimpse of what Mark Borh said in a interview some time ago:

http://semimd.com/blog/2012/10/18/deep-inside-intel/

Btw, that interview is full of hidden nuggets. It's been a year, but it's a worthy reading.

It pretty much says what most of us tries to surpress in our minds. That desktops is a dying breed in favour of mobile devices of all sorts, be it laptops, tablets, phablets, smartphones or whatever name there is.

And we still expect Intel to cater for the niche market that desktop has become as a first priority. It was, as you write, exactly that mistake IBM essentially made. Betting on the dinosaur with the mainframes.

You can always use a mobile CPU in a desktop. But you cant use a desktop CPU in a mobile device.

 

[jvroig]

Actually, If I remember correctly those weren't his words at all. He characterized overclocked Haswell to be "interesting" due to the new bclk overclocking feature, but didn't really go further than that. Of course everyone interpreted that to mean Haswell being what Sandy Bridge was a few years back.

No, you don't remember correctly. That was an actual quote. Here. Look at Item #3.

He said worth it, look for buried treasure, greatly imparting the meaning that "yep, gonna be sweet for all you overclockers". I distinctly remember it because it is what got me to hope greatly about Haswell OC. And then we have reality, which is pretty much the opposite, because there is zero boon for overclockers.

 

[BallaTheFeared]

Actually, If I remember correctly those weren't his words at all. He characterized overclocked Haswell to be "interesting" due to the new bclk overclocking feature, but didn't really go further than that. Of course everyone interpreted that to mean Haswell being what Sandy Bridge was a few years back.

That said, people obviously expected a little bit more with Haswell overclocking - The desktop user in me wanted more, I don't disagree there at all. As things are, though, it's still an appreciable IPC increase over IVB which will offset any potentially lower OC anyway compared to SB or IVB.

Intel is moving forward at pretty much the same clip they have been since Conroe.

Everyone has a 4.2GHz 920, but most of them were doing 3.4-3.6GHz back in the day before steppings improved the clocks. You know there were Q9xxx users with 4GHz chips equally unimpressed, if it wasn't for HT it would have been even worse.

I don't understand the doom and gloom here though, were people really expecting clock speeds to continue to go up well past 5GHz? Haven't netburst and bulldozer shown that's the wrong way to go already? Maybe TSX will be a benefit for performance to the market that generally overclocks, maybe it won't... Until it does why worry about it?

 

[cytg111]

It pretty much says what most of us tries to surpress in our minds. That desktops is a dying breed in favour of mobile devices of all sorts, be it laptops, tablets, phablets, smartphones or whatever name there is.

And we still expect Intel to cater for the niche market that desktop has become as a first priority. It was, as you write, exactly that mistake IBM essentially made. Betting on the dinosaur with the mainframes.

You can always use a mobile CPU in a desktop. But you cant use a desktop CPU in a mobile device.

- For good? I can see a segment that needs filling, but IMO there is only so much you can do with a mobile device(tablet/phone), to truely enable yourself you're gonna need something desktop like. That might be a 4K tellie with kinnect like interfaces, but still, a box is needed to push those 4K pixels at 3D and beyond.
I see the desktop taking a nap .. It will be back.

 

[NTMBK]

Derp! You are right, my bad



There once was a time when Intel was the Calxeda of the microprocessor market.

At the time there were companies who had already been dominating the space for decades already, who had nice fat margins and a performance lead that was second to no other.

And then Intel came along, as scrappy as it was at the time, and completely took their lunch money and put nearly every single one of them out of business.

Now here we are, 20-30 yrs later, a whole new generation of lethargic and lazy management is in control (hell they were just kids when Intel's management was laying the groundwork for becoming an 800lb gorilla)...and today's management are doing exactly the same thing that DEC's management did to set the foundation for yet another cycle of smaller scrappier companies to come along and eat their lunch.

I'm not at all worried about who exactly is going to be "the next Intel", that is for wallstreet speculators to fret over predicting, but I am absolutely comfortable with placing my bets on history repeating itself and Intel's complacency and managerial hierarchy being their own worst enemy.



Said a Fairchild semiconductor exec circa 1974....and a Cray exec circa 1984...and a DEC corporate exec circa 1994...and a SUN exec some 10yrs later...and an Intel exec 10yrs later still.

Go back in recorded history as far as you like and one trait is shared 100% across all business entities regardless of nationality and continent - eventually complacency and hubris combine forces and the outcome is the same over and over again.

Intel is clearly no longer paranoid in the desktop and server markets, and only the paranoid survive.

I think Intel is aware of the parallels, and their recent refocus on mobile is a sign that they are trying to avoid the same fate. They're finally bumping Atom up to the leading edge and bringing it out on 14nm within the process' first year (instead of lagging behind, as standard).

The risks to them is from disruptive "good enough" cores in vast numbers- whether it is a phalanx of ARM microservers, or a few GPGPUs. Their dedication to bringing the Xeon Phi to market, as well as making Atom more aggressive once again, seem like promising signs. I just hope that they can learn to suck up lower margins, and stop playing these damn market segmentation games.

 

[NTMBK]

- For good? I can see a segment that needs filling, but IMO there is only so much you can do with a mobile device(tablet/phone), to truely enable yourself you're gonna need something desktop like. That might be a 4K tellie with kinnect like interfaces, but still, a box is needed to push those 4K pixels at 3D and beyond.
I see the desktop taking a nap .. It will be back.

As others have said before me, I suspect that desktop is just seeing a readjustment- not a death. Those who really need a large box for powerful parts chomping on a lot of energy will never give them up- basically the current workstation market, and gaming PC markets. But does browsing Facebook and watching Netflix really need more processing power than a less powerful, more compact machine can offer?

 

[ShintaiDK]

- For good? I can see a segment that needs filling, but IMO there is only so much you can do with a mobile device(tablet/phone), to truely enable yourself you're gonna need something desktop like. That might be a 4K tellie with kinnect like interfaces, but still, a box is needed to push those 4K pixels at 3D and beyond.
I see the desktop taking a nap .. It will be back.

I would say that I see future desktop boxes with mobile chips instead of dedicated desktop chips. There will still be workstations with servergrade chips.

 

[cytg111]

As others have said before me, I suspect that desktop is just seeing a readjustment- not a death. Those who really need a large box for powerful parts chomping on a lot of energy will never give them up- basically the current workstation market, and gaming PC markets. But does browsing Facebook and watching Netflix really need more processing power than a less powerful, more compact machine can offer?

Nope, but that same box could easily drive a homes total compute power and then you got your small compute(tablet/phone) when mobile.
I know that I would love to do my work from the couch lying down, waving my hands around and speaking to my "'desktop".. I have no issue with the desktop moving out of my office, quite the opposite.

There is another angle too, right now the x86 market is flooded with 4 to 8 core parts and while the software industry at large is getting into harvesting that parallel potential, it makes good sense to focus on the watts, suppose you were to put 4 bulldozers at 150 watts into a box vs 4 haswells at 65 ... One of them is doable, the other is not. So focusing on power now may well be a bet on maximum computing power for the future.

 

[Idontcare]

As others have said before me, I suspect that desktop is just seeing a readjustment- not a death. Those who really need a large box for powerful parts chomping on a lot of energy will never give them up- basically the current workstation market, and gaming PC markets. But does browsing Facebook and watching Netflix really need more processing power than a less powerful, more compact machine can offer?

Nothing dies, there is still a market for horse buggies and the necessary horse whips.

So saying desktop won't die isn't really making a strong stand because no one is really saying its death is coming.

But we all know what happens once a form factor or product falls out of favor. It stops getting priority, exposure, and R&D begins to dry up.

Very little R&D is going into developing better and better vinyl records, landline telephones, and leaded-gasoline. But you can still buy all of those products, as stagnate as they are.

That is where the desktop is headed, eventually, IMO. There will always be a need for them, but the market volumes that represent that base need are going to be silly small. Engineering stations, design, etc.

But things have peaked, as they inevitably do, and that makes other things become the priority for today's R&D dollar.

IBM, SUN and HP show that dedicated server CPUs have a place for low-volume high-performance server markets. That market isn't going away just because mainstream consumer PCs are in a decline.

But in my opinion the consumer desktop is this decades landline, and it is falling out of favor big-time with this decades kids who are going to be young adults in 10yrs and won't give a damn about buying a desktop (just as I haven't paid for a landline these past 6 yrs).

For the 5-6% of them that go on to become engineers they will still have a desktop at work, the other 95% will be on something mobile.

 

[cytg111]

That is where the desktop is headed, eventually, IMO. There will always be a need for them, but the market volumes that represent that base need are going to be silly small. Engineering stations, design, etc.

But things have peaked, as they inevitably do, and that makes other things become the priority for today's R&D dollar.

- I dont see it. The need for total computing power will always be on the incline, driven by games, porn or otherwise virtual realities enabling people in their day to day tasks.
Fx. imagine working from home. Now imagine that you dont have to imagine.
But thats just my personal crystal ball.

 

[Vesku]

- I dont see it. The need for total computing power will always be on the incline, driven by games, porn or otherwise virtual realities enabling people in their day to day tasks.
Fx. imagine working from home. Now imagine that you dont have to imagine.
But thats just my personal crystal ball.

Yes but companies now want to go back to the rental model with the "cloud". So their goal is that eventually the fastest chips are out of reach of most individuals and mainly in their servers.