Is Desktop Haswell a dud? How many are not "upgrading"?

[Durvelle27]

i'm not getting it


XD

 

[CakeMonster]

How many threads with the exact same topic do we need?

 

[Ajay]

Haswell-E will be the Haswell chip worth buying.

I'm betting on that as well. I hope we are right.

LOL.. well yes. More like the end of 2014 though.. hopefully.

The date seems to keep bouncing around between late 2014 and the first half of 2015. This may just be made up stuff for page hits, or it could be Intel's new management shuffling some things around. We probably won't get reliable info till after IVB-E is out.

 

[Piano Man]

For the desktop enthusiast, its a bomb. I have the 3770k, but even if I had 2700k, I wouldn't even think of upgrading.

Not like I expect it to happen but I wish AMD could do something about this. This is what happens when Intel has no competition.

 

[gammaray]

i got a 3570k home and now "my" 4670 is on the way home!

why not?

 

[njdevilsfan87]

It makes no sense to get one. You're better off either getting a used SB or new IB combined with a heavily discounted motherboard from Amazon Warehouse or Newegg Open Box. I recently got a new i5-3570K second hand and like new ASUS Z68 Pro Gen/3 from Amazon for $270 combined. A 4670K would run over $400 for a similar setup.

And this is coming from someone who spent $1000+ on a Titan. I'm truly disappointed in this release. :sneaky:

 

[gammaray]

I recently got a new i5-3570K second hand and like new ASUS Z68 Pro Gen/3 from Amazon for $270 combined. A 4670K would run over $400 for a similar setup.

you could say the same thing about cars, makes no sense to get a brand new one, when last year used ones are 30% cheaper.

when you want something newly released, you gotta pay.

 

[BallaTheFeared]

I just look at it as a new toy with far more tweaking options than Sandy Bridge.

I found SB to be crazy fast, but stupidly boring at the same time. Hopefully I can have a bit more fun with Haswell, but if not oh well I won't lose sleep over it.

Steps to overclocking SB:

Step 1: Enter Voltage

Step 2: Set Turbo

Step 3: Validate

 

[aigomorla]

Over the years, I have always liked your posts. And you seem very knowledgeable and experienced. But this is the dumbest thing I have ever heard you say.

Intel succeeded in their goal (Mobile). And their new instruction sets are huge. Just because it is not the Uber gaming CPU, everyone is all bent out of shape. There is more to CPUs than gaming and OC records.

I am talking about the haswell desktop.. not mobile.
I know of all the mobile accomplishments, however were not looking for mobile on desktops.

I say they pulled a bulldozer, because like the Athlon -> Bulldozer migration, were seeing the same results on the intel scale.

We got a slightly faster chip.. nothing to go WOW about... with a LOT more heat... exactly like how bulldozer was when first released.

A long time ago... well not so long time ago... maybe 3 yrs ago.. a lga1160 was identical in most regards to a LGA1366 chip.
Before that LGA771 was identical to LGA775.

Once we started rolling LGA1155 vs LGA2011... eh.. something went seriously wrong...

Now... were getting a LGA1150...
Dont get me wrong... when i talked to a guy at intel, they said they hated making board companys change sockets... LGA1155 was entirely to fix a problem...
So are you telling me the LGA1155 has a serious problem in design which merited it a entire board revision?

Oh yeah its so we could get VRMS on the cpu directly, when boards have been doing a very good job at it already...
^this makes WAY MORE SENSE on a mobile or ultra portable scale. Not on a big desktop which sits under a desk..

Did we really NEED those VRMs on the cpu on a DESKTOP?
Did we really NEED a new socket on a platform which is less then 2 yrs old?

What happened to my rep friend who told me intel doesnt want us to change sockets so quickly?
What have desktop people been doing since LGA1160?

So anyhow on with my rant...

It all branched from LGA775 -> went two ways... LGA1160 vs LGA1366
Then LGA1160 -> LGA1155 -> LGA1150
While LGA1366 -> LGA2011 -> LGA20XX

Yeah intel sure doesnt want us to change boards.. :\

For the desktop enthusiast, its a bomb. I have the 3770k, but even if I had 2700k, I wouldn't even think of upgrading.

Not like I expect it to happen but I wish AMD could do something about this. This is what happens when Intel has no competition.

sigh... i am very disappointed in haswell at the moment... because i was really expecting to upgrade my aging system.

i got a 3570k home and now "my" 4670 is on the way home!

why not?

Have fun with the earily board issues lga1150 is running into.
I hear its a repeat of LGA1160 issues all over again.

People are reporting issues with USB3.0 and all other stuff.. :\

This is why i really tell people to wait a bit before you buy.
Ive ran into problems where retail boards were bugging out more then the ES boards because of a bios revision which barf'd the board even more.

So i can tell being an earily adopter has about 1-2 weeks of glory... however for those 1-2 weeks, you can spend 1 - 2 months debugging, and still not know whats exactly wrong with your PC.

 

[BallaTheFeared]

I am talking about the haswell desktop.. not mobile.
I know of all the mobile accomplishments, however were not looking for mobile on desktops.

I say they pulled a bulldozer, because like the Athlon -> Bulldozer migration, were seeing the same results on the intel scale.

We got a slightly faster chip.. nothing to go WOW about... with a LOT more heat... exactly like how bulldozer was when first released.

A long time ago... well not so long time ago... maybe 3 yrs ago.. a lga1160 was identical in most regards to a LGA1366 chip.
Before that LGA771 was identical to LGA775.

Once we started rolling LGA1155 vs LGA2011... eh.. something went seriously wrong...

Now... a LGA1155 cpu... IS NOT THE SAME as a LGA2011.

By the same token Ivy was a fail as well. (Little IPC gain, huge increase in chip temps)

Your comparisons are one off as well.

Bulldozer put out a lot of heat because it used a lot of power, Haswell does not. It only runs hot because of the thermal interface, it doesn't actually use a lot of power.

Bulldozer was a serious loss of IPC with little in the way of MHz gain over Phenom II.

Intel separated mainstream from server/enthusiasts. They push newer tech out the mainstream pipe, which is great for almost everyone here because not many of us need the benefits of the high end socket.


If Intel didn't drop some of the advances in Haswell into Ivy Bridge than Ivy wouldn't have had the IPC gains it had, and which case Ivy would have been the same as the 45 > 32 nm products last time (no IPC gain), Haswell is still bringing decent increases in perf/w and IPC over Sand Bridge, it just used Ivy to validate some of those changes no big deal.

In fact if you compare Bloomfield to SB to Hawell the IPC increases are nearly the same.

 

[philipma1957]

looks to me like the i5 4570s makes sense


http://ark.intel.com/products/75044/Intel-Core-i5-4570S-Processor-6M-Cache-up-to-3_60-GHz

as a silent ht chip it has better graphics then the i5 3470s hd2500 or the i5 3570s hd2500 if you go with no gpu price is okay.

http://ark.intel.com/products/65701/Intel-Core-i5-3570S-Processor-6M-Cache-up-to-3_80-GHz

http://ark.intel.com/products/68315/Intel-Core-i5-3470S-Processor-6M-Cache-up-to-3_60-GHz

 

[njdevilsfan87]

Yeah Haswell is definitely no Bulldozer. The 15% IPC gain over SB is... fair... it's just that SB clocks so well that in order to see that gain you'd need to get a Haswell. If I had a crappy clocking SB or IB chip, I'd probably give Haswell a shot.

 

[Wall Street]

Just to turn the clock back a little bit:

Why Intel's new CPUs disappoint

Seems like this sentiment is almost an annual occurance. I can find artlicles about Sandy Bridge being a disappointment too since Lynnfield (which had BClk adjsustments) could overclock to 2500k performance.

 

[BallaTheFeared]

^

Indeed :thumbsup:

 

[njdevilsfan87]

I don't think SB was disappointing. It made for small IPC increase through being a monster overclocker (which increased the performance gap between it and Nehalem to 20-25%), and did it while consuming much less power, running cool, and costing less for the entire LGA 1155 platform (vs X58). In fact, the reason Haswell is disappointing is pretty much because of SB, and probably the 2500K specifically. The first generation can argue that as well (i7 920 D0), but those weren't anywhere near as popular as the 2500K.

 

[aigomorla]

By the same token Ivy was a fail as well. (Little IPC gain, huge increase in chip temps)

But you see... at least IVY didnt require a new board!!!

 

[Ayah]

Just to turn the clock back a little bit:

Why Intel's new CPUs disappoint

Seems like this sentiment is almost an annual occurance. I can find artlicles about Sandy Bridge being a disappointment too since Lynnfield (which had BClk adjsustments) could overclock to 2500k performance.

No competition = No need to bother innovating. Just a small improvement is all you need.

If AMD had produced another A64 vs P4 performance differential to light a fire under Intel's six, the playing field might've been different and we'd all be "oooh, sex on silicon".

 

[BallaTheFeared]

I don't think SB was disappointing. It made for small IPC increase through being a monster overclocker (which increased the performance gap between it and Nehalem to 20-25%), and did it while consuming much less power, running cool, and costing less for the entire LGA 1155 platform (vs X58).

Well there was Lynnfield.

But when I was running my i5-2500k @ 5.3GHz there were still people with those older core chips saying "It isn't worth it" "I expected more".

And that wasn't just people with core i systems, that was Core2 and Core2quad users...

Without the clock speed increase, Haswell isn't as good as SB was imo. However I know a lot of SB users that couldn't get much over 4.4GHz, there are always going to be people with golden samples who get high clocks with low voltage.

Even without the clock speed advantage it still has an IPC and instruction set advantage, we'll see how it plays out, but if SB has the MHz advantage, it sure didn't have an instruction set one over previous chips... Perhaps that will be the x factor for Haswell, either way it's by no means a "fail" it's just hamstrung.

 

[aigomorla]

If AMD had produced another A64 vs P4 performance differential to light a fire under Intel's six, the playing field might've been different and we'd all be "oooh, sex on silicon".

its totally this..

The A64 was such a tough chip for intel, they needed to slaughter it 3 generations over...

As it got slaugheterd... C2D -> i7 Bloomfield -> i7 Sandy they realized they are so far ahead.. they are now getting sloppy again...

Its like Intel is trying to throw netburst at us again, and seeing if we will look the other way. :\

 

[Edrick]

Once code is written to take advantage of AVX2/FMA, you all will change your tune about Haswell. Sure, it will take some time, but Intel had to start somewhere, and they had to be first (before developers).

Lets try and being realistic here. There is only so much more juice Intel can squeeze out of legacy x86 code without new instructions.

 

[FwFred]

Ivy Bridge E (E5-2600 v2 family) is Q3... for those who want MOAR CORES can get 12 per socket for about the same TDP as Piledriver FX-8350

Tom's article on Ivy Bridge EP

 

[aigomorla]

Ivy Bridge E (E5-2600 v2 family) is Q3... for those who want MOAR CORES can get 12 per socket for about the same TDP as Piledriver FX-8350

Tom's article on Ivy Bridge EP

those cpu's are on a completely different socket then 2011 tho no?
If its not, then...

Tim Allen Grunt..
MOAR CORES!!!

 

[FwFred]

those cpu's are on a completely different socket then 2011 tho no?
If its not, then...

Tim Allen Grunt..
MOAR CORES!!!

Ivy Bridge E compatible with LGA 2011

Edit: Now that I think about it, I'm not sure if Xeons can just pop in to a i7 extreme motherboard. Probably very BIOS dependent.

 

[rgallant]

How many threads with the exact same topic do we need?

8 threads

 

[VirtualLarry]

Bulldozer put out a lot of heat because it used a lot of power, Haswell does not. It only runs hot because of the thermal interface, it doesn't actually use a lot of power.

I wouldn't be so sure about that. Doesn't Haswell's FIVR basically overvolt when you OC the K chips, like old mobo's used to overvolt CPUs when they were set to AUTO voltage?

Except now with FIVR, we the end-user no longer have any sort of manual control over the multitude of voltage rails within Haswell. (I would love to be proven wrong about this.)

TLDR: Basically, we're back to the bad old days of AUTO vcore overclocking.