Today, we can now buy Haswell processors, like the 4770k. TomsHardware has shown them to be more like a 3870k, with no real massive improvements over the 3770k, increased power consumption, and lessoverclocking potential. On the other hand, Z87 looks nice, as it offers more PCIe lanes and USB ports.
Out only hope is for AMD to get its act together, now. What do you guys think?
i5 4670k @ 4.9GHz - Stock Heatsink - The rest is melted silicon but I think I have a graphics card in there somewhere It surprises me how many people on this forum can't read benchmarks.
Congrats Intel you can nearly match AMD Trinity.
While this would be a nice improvment if it was not for the fact Richland is coming out this month and Kaveri will be out at the end of the year.
It has some single thread performance increase it can stay at turbo longer but overall its meh.
I think I like the motherboards this time around much more than the Z77 series.
Now that MSI and Gigbyte are both in the red and black theme, I don't have to get screwed over by Asus to get a red and black motheboard. The Z77 Maximus series are like 300$ at least.
I'm not really surprised by this considering Intel is focusing more on the mobile market. What I am surprised by is the higher power draw and worse overclocking ability. Haswell was supposed to be an overclocking monster, with rumours that suggested 5GHz overclocks would be easily attainable. At this point, I'm more excited for AMDs Kaveri APU. Steamroller architecture improvements, a new motherboard socket FM2+ with either integrated GDDR5 or a slot for adding GDDR5, and iGPU performance rumoured to be between the HD7770 and 7790. A top-of-the-line Kaveri APU could be a suitable replacement for my aging Phenom II and HD6850.
AMD never really go lazy, they just had poor management. Leaked SR module shots look quite promising.
Piledriver underwhelmed everyone just like the 4770k, Bulldozer was and still is able to throw punches at intel's mid/high end (3570k/3770k). Overall, if AMD can live up to their predicted performance increases I am quite optimistic about steamroller.Also, link to the steamroller leak?
I also thought Bulldozer was a bit of a flop and Piledriver was a win. Anyway I wasn't expecting great gains in terms of Haswell, especially for us desktop users.
My bad, I meant to say that Piledriver was and is able to compete with the 3570k and the 3770k.
Somehow puts out more heat than SB or IB, even with less power consumption
This is due to the VRMs now being on-die vs on the mainboard. Increases the temps by about 10C.
Can barely be overclocked without throttling due to heat
Does not mean you won't be able to overclock, it just means you will have to pick up an aftermarket heatsink to overclock, and will need to build a custom loop to go over voltages of about 1.25.
Integrated voltage controller that throttles extremely close to stock
Again, heat issue, if you want to overclock get an aftermarket cooler.
Some overclockers able to get 5Ghz idle but any load crashes the system instantly
4.8Ghz at 1.25v is decent for a good chip, so under a good water loop, 1.3v at 5Ghz stable seems doable.
I hear that single-stage/phase-change cooling is going to be making a comeback for Haswell.
Personally, I'm not underwhelmed, it's simply a "tock" vs a "tick". Innovation vs performance increases. That being said, I'm waiting for the next "tick" before I upgrade. I won't have the money or need to upgrade until then.
[EDIT]
Quote from Overclockers Forums »
With Haswell, you’re going to see a big push for energy efficiency. The name of the game right now is mobile. While I like to think the desktop (especially the enthusiast desktop) market is still here for showing the best of the best, and what these CPUs are capable of when you build a balls-the-the-wall system around them. As you can see, Intel did tweak quite a bit with Haswell and, while not earth shattering, should still give some noticeable performance benefits. However, while there is increased performance, energy efficiency appears to have been their main focus.
i5 4670k @ 4.9GHz - Stock Heatsink - The rest is melted silicon but I think I have a graphics card in there somewhere It surprises me how many people on this forum can't read benchmarks.
before i do anything with them (suggest or anything else) there needs to be more motherboards made...
Rollback Post to RevisionRollBack
Proud member of the MCF AWA war of '13! if someone suggests Alienware or Cyberpower, wait for a custom-built list from someone who knows their stuff. Meh Rig
This is due to the VRMs now being on-die vs on the mainboard. Increases the temps by about 10C.
Does not mean you won't be able to overclock, it just means you will have to pick up an aftermarket heatsink to overclock, and will need to build a custom loop to go over voltages of about 1.25.
4.8Ghz at 1.25v is decent for a good chip, so under a good water loop, 1.3v at 5Ghz stable seems doable.
[EDIT]
Overclockers Forums said: With Haswell, you’re going to see a big push for energy efficiency. The name of the game right now is mobile. While I like to think the desktop (especially the enthusiast desktop) market is still here for showing the best of the best, and what these CPUs are capable of when you build a balls-the-the-wall system around them. As you can see, Intel did tweak quite a bit with Haswell and, while not earth shattering, should still give some noticeable performance benefits. However, while there is increased performance, energy efficiency appears to have been their main focus.
It is a flop for now, nuff said.
What idiots in their wrong mind, note in their wrong mind, would change something that worked fine. What i talk about; what idiot would put VRMs on a device that has never had issues with voltage regulation anyway from the motherboard itself. Stupid move if anything, quite very stupid.
While my opinion, it should remain so, but still quite a dumb move. If it is not causing problems nor is it broken, it do not need fixing or replaced.
Custom loop to go over 1.25 volts? Is this Sparta? No, this is just WTF. However yes, such is kinda a misnomer or I hope it to be so.
Sad that SB did easy go to 5Ghz on a good air cooler, sometimes that of the Hyper 212+ Evo with decent temperatures. Ivy bridge was nope really and now Haswell not even near such. Intel seems to be neutering peoples abilities to overclock and regain control. Gee, someone Red, Black and Green is looking mighty tasty for extreme overclockers or overclockers in general again.
Did you really quote something from Overclockers? They say anything at moment to try and get spotlight on them by Intel or anyone. Then they cry in a corner because they cannot overclock and flex epeens anymore with Haswell due to heat. That place is so sadly pitiful, pitiful enough that it isn't worth to discuss at all about it.
Remember those are opinions and are kept such.
Now that such is done.
I really am quite displeased with the views of Haswell at moment. While amused that it just now has applied to Moore's Law, given 3 years transistor count doubles now instead 2 years. Lynnfield which was roughly 3 years ago, was 774mil transistors and now Haswell which is 1.4 to 1.6 billion transistors. But that is 3 years apart at most.
However now it gets a bit sad, the cost of computer power for an consumer is increased a bit because this dumb move about VRM on chip ideal. But even then this is negated by Rock's Law.
Intel is in the cat and mouse game it seems. But if the Cat gets smart enough, it will sit and be silent then pounce and play with the mouse before killing it or letting it go for another day.
This is due to the VRMs now being on-die vs on the mainboard. Increases the temps by about 10C.
Does not mean you won't be able to overclock, it just means you will have to pick up an aftermarket heatsink to overclock, and will need to build a custom loop to go over voltages of about 1.25.
Again, heat issue, if you want to overclock get an aftermarket cooler.
You are missing the part where these overclocks I am referencing were on custom loops with the heat spreader and crappy silicon based TIM removed.
4.8Ghz at 1.25v is decent for a good chip, so under a good water loop, 1.3v at 5Ghz stable seems doable.
Again, thats what these tests were. Good chips, custom loop, heat spreader removed, 5GHz idle was the best they could do, any load would crash the system.
Personally, I'm not underwhelmed, it's simply a "tock" vs a "tick". Innovation vs performance increases. That being said, I'm waiting for the next "tick" before I upgrade. I won't have the money or need to upgrade until then.
The issue is this was supposed to be a "tick", not a "tock" as it is.
Core 2 was a "tick"
Core i# was a "tock"
Sandy was a "tick"
Ivy was a "tock"
Haswell should have been another "tick"
You are missing the part where these overclocks I am referencing were on custom loops with the heat spreader and crappy silicon based TIM removed.
Again, thats what these tests were. Good chips, custom loop, heat spreader removed, 5GHz idle was the best they could do, any load would crash the system.
The issue is this was supposed to be a "tick", not a "tock" as it is.
Core 2 was a "tick"
Core i# was a "tock"
Sandy was a "tick"
Ivy was a "tock"
Haswell should have been another "tick"
Both Ivy and Haswell might as well have been a 'tick'.
Both Ivy and Haswell might as well have been a 'tick'.
what kind of clock ticks twice and tocks once?
Rollback Post to RevisionRollBack
Proud member of the MCF AWA war of '13! if someone suggests Alienware or Cyberpower, wait for a custom-built list from someone who knows their stuff. Meh Rig
Both Ivy and Haswell might as well have been a 'tick'.
Err, how? A tick implies a new level of performance and engineering, a "tock" is only a minor improvement overall.
Sandy was a tick, Ivy/haswell are not. Ivy's performance was only slightly higher than sandy, and power consumption was really all that was improved.
Haswell's power consumption was improved, but yet again its performance is similar to both sandy and ivy (worse than ivy and sandy in some areas) and due to it giving off more heat, it's less of an improvement in the areas than it is just a minor sidegrade.
Thanks for posting SM8. Its a pitty the processor has not gotten significantly better except for the GPU, however I can smile knowing my recent purchase is not going to be super-seeded any time soon.
Out only hope is for AMD to get its act together, now. What do you guys think?
http://www.tomshardw...eview,3521.html
It surprises me how many people on this forum can't read benchmarks.
While this would be a nice improvment if it was not for the fact Richland is coming out this month and Kaveri will be out at the end of the year.
It has some single thread performance increase it can stay at turbo longer but overall its meh.
Now that MSI and Gigbyte are both in the red and black theme, I don't have to get screwed over by Asus to get a red and black motheboard. The Z77 Maximus series are like 300$ at least.
Piledriver underwhelmed everyone just like the 4770k, Bulldozer was and still is able to throw punches at intel's mid/high end (3570k/3770k). Overall, if AMD can live up to their predicted performance increases I am quite optimistic about steamroller.Also, link to the steamroller leak?
Cool. So this long of a development time we should see some nice improvements like---
........... how?
Was this not supposed to be a new "cycle" in the 'tick, tock, tick, tock' process?
At least the new mobo features are nice, but not warranting something like this for a cpu.
My bad, I meant to say that Piledriver was and is able to compete with the 3570k and the 3770k.
This is due to the VRMs now being on-die vs on the mainboard. Increases the temps by about 10C.
Does not mean you won't be able to overclock, it just means you will have to pick up an aftermarket heatsink to overclock, and will need to build a custom loop to go over voltages of about 1.25.
Again, heat issue, if you want to overclock get an aftermarket cooler.
4.8Ghz at 1.25v is decent for a good chip, so under a good water loop, 1.3v at 5Ghz stable seems doable.
I hear that single-stage/phase-change cooling is going to be making a comeback for Haswell.
Personally, I'm not underwhelmed, it's simply a "tock" vs a "tick". Innovation vs performance increases. That being said, I'm waiting for the next "tick" before I upgrade. I won't have the money or need to upgrade until then.
[EDIT]
EDIT: I have also updated my signature to account for Haswell's limited overclocking capabilities
It surprises me how many people on this forum can't read benchmarks.
Thinking about coming a mod to simply not moderate.
if someone suggests Alienware or Cyberpower, wait for a custom-built list from someone who knows their stuff. Meh Rig
It is a flop for now, nuff said.
What idiots in their wrong mind, note in their wrong mind, would change something that worked fine. What i talk about; what idiot would put VRMs on a device that has never had issues with voltage regulation anyway from the motherboard itself. Stupid move if anything, quite very stupid.
While my opinion, it should remain so, but still quite a dumb move. If it is not causing problems nor is it broken, it do not need fixing or replaced.
Custom loop to go over 1.25 volts? Is this Sparta? No, this is just WTF. However yes, such is kinda a misnomer or I hope it to be so.
Sad that SB did easy go to 5Ghz on a good air cooler, sometimes that of the Hyper 212+ Evo with decent temperatures. Ivy bridge was nope really and now Haswell not even near such. Intel seems to be neutering peoples abilities to overclock and regain control. Gee, someone Red, Black and Green is looking mighty tasty for extreme overclockers or overclockers in general again.
Did you really quote something from Overclockers? They say anything at moment to try and get spotlight on them by Intel or anyone. Then they cry in a corner because they cannot overclock and flex epeens anymore with Haswell due to heat. That place is so sadly pitiful, pitiful enough that it isn't worth to discuss at all about it.
Remember those are opinions and are kept such.
Now that such is done.
I really am quite displeased with the views of Haswell at moment. While amused that it just now has applied to Moore's Law, given 3 years transistor count doubles now instead 2 years. Lynnfield which was roughly 3 years ago, was 774mil transistors and now Haswell which is 1.4 to 1.6 billion transistors. But that is 3 years apart at most.
However now it gets a bit sad, the cost of computer power for an consumer is increased a bit because this dumb move about VRM on chip ideal. But even then this is negated by Rock's Law.
Intel is in the cat and mouse game it seems. But if the Cat gets smart enough, it will sit and be silent then pounce and play with the mouse before killing it or letting it go for another day.
Intel is now the mouse and AMD is the cat.
Let the game begin.
Again, thats what these tests were. Good chips, custom loop, heat spreader removed, 5GHz idle was the best they could do, any load would crash the system.
The issue is this was supposed to be a "tick", not a "tock" as it is.
Core 2 was a "tick"
Core i# was a "tock"
Sandy was a "tick"
Ivy was a "tock"
Haswell should have been another "tick"
Both Ivy and Haswell might as well have been a 'tick'.
what kind of clock ticks twice and tocks once?
if someone suggests Alienware or Cyberpower, wait for a custom-built list from someone who knows their stuff. Meh Rig
Sandy was a tick, Ivy/haswell are not. Ivy's performance was only slightly higher than sandy, and power consumption was really all that was improved.
Haswell's power consumption was improved, but yet again its performance is similar to both sandy and ivy (worse than ivy and sandy in some areas) and due to it giving off more heat, it's less of an improvement in the areas than it is just a minor sidegrade.
Looks like a pretty crappy 'upgrade'. Guess I'm keeping my 2500k for at least another year or two.