Should I wait for Haswell?

Arbegust

Distinguished
Nov 6, 2011
54
0
18,630
I want to upgrade my PC in the next six months, mainly to future proof for gaming. I don't have a bad rig now, but would like PCI 3 capacity. Therefore my question is should I wait for Haswell or purchase a Sandy Bridge E or Ivy Bridge? My specs are as follows:

Intel i5 2500k 3.3Ghz
Nvidia GTX 680
12 gig DDR 3 RAM
750w Corsair PSU
1 TB WD HDD

-Oh yes, I mainly play MMOs and am looking forward to the Elder Scrolls. Also, do I need DDR 4 RAM?

Thanks so much for any help!
 

neon neophyte

Splendid
BANNED
neither sandybridge e or ivybridge would be anymore futureproof for gaming. 2500k is still top notch. i would overclock it though.

when overclocked sandybridge is just as good as ivybridge (as it can overclock more than ivybridge and make up for any inferior ipc it has.)
 
Sandy Bridge is a great processor! It handles everything really well and is very powerful still; it overclocks great too.

Wait till Broadwell, it's 14nm, Haswell will only be 22nm just like Ivy-Bridge. I bet we will be pushing 4.8ghz and higher easily on air coolers.


960px-IntelProcessorRoadmap-3.svg.png

 

andrewcarr

Distinguished
I thought you'd have something like an i3. With that processor theirs no reason to upgrade when you'd only see a ~5% improvement getting an ivy bridge and sandy bridge-E just isn't worth it if all you want to do is game.
Conclusion (as I'm doing before building a new computer) is to wait for Haswell, although I'm still not sure you'd need to upgrade then.
 

openfire070

Honorable
Sep 3, 2012
44
0
10,530
I thought you'd have something like an i3. With that processor theirs no reason to upgrade when you'd only see a ~5% improvement getting an ivy bridge and sandy bridge-E just isn't worth it if all you want to do is game.
Conclusion (as I'm doing before building a new computer) is to wait for Haswell, although I'm still not sure you'd need to upgrade then.

I so agree ^
 
Yeah, the 2500k will be good for at least the next couple of years. No single GPU uses more than slightly over half the bandwidth of PCI-E x16 2.0, so getting a new platform for PCI-E 3.0 is pointless right now. I'd say wait until Broadwell in 2014 at the earliest before you look at upgrading. You probably will be good until Skylake in 2015 if all you are doing is gaming.
 

matt_b

Distinguished
Jan 8, 2009
653
0
19,010


You know, I could totally agree with this statement if it weren't for the issues we ran in to with Ivy Bridge. It's no secret now that Intel cheaped out on the heat spreader application by using TIM instead of flux-less solder, but will Haswell and Broadwell have the same application and problem associated with it as Ivy does? Furthermore, 22nm brought about a problem that the solution to has yet to be found, the die surface area. It has gotten to the point now that it is so small that we cannot transfer enough heat fast enough to the heat-spreader to get rid of the heat at a given thermal load. What this means is that further down the road, the "wall" overclocking presents will hit even harder as a chip will reach the point where temps will absolutely take off (see how poor Ivy handles this). I'm trying to imagine how the 14nm die shrink will behave, some pretty innovative ideas are going to have to flow out of the engineering department to overcome this obstacle. So, 4.8ghz easy, I'm not betting on it if it's the progression of the 2500k ->3570k -> xxxxK series and they are constructed like Ivy is. As it is now, anything past 4.5 ghz for a 3570k can be difficult to acheive.

If trying to buy in the next 6 months, just pick the best deal you can period. We are already at a point where most processors just aren't taxed to their fullest unless a very high-end multi-GPU setup is used. Haswell is shaping up to be more power efficiency and a good boost to the IGPU, that's about it though - some even are questioning if it's a "tick" or a "tock" by the sound of it. By waiting, you're probably not gaining a whole lot here by doing so.

DDR 4 is still a ways off, at least a year last I heard. Don't expect this stuff to be cheap at launch by any means either.

 



That is a good question if they will use TIM or a fluxless solder in the next line of processors IHS; Ivy-bridge would have been very great if it wasn't for that switch. I doubt they are really saving that much in profits by using a TIM; I would rather pay the cost for them to just use solder.

14nm should be very power efficient and be a large upgrade over 22nm which is why I'm waiting for the 14nm CPUs. My 2500k runs cooler at 4.5ghz compared to my old e6600 I had ages ago at 3.2ghz(both using good heatsinks), I always expect good overclocking potential on a lower NM. I honestly can't imagine my 2500k not running any game that comes out in the next 3 years so 2014 will be a good year for an upgrade and DDR 4 should be cheaper and more standard by then too.
 
There is no need to upgrade to Ivy Bridge since the average performance increase assuming the same clock speed is only about 5% or 6%. That will not significantly improve gaming performance.

There's not much certainty about how much of an improvement Haswell will be over Ivy Bridge. I figured it would be something between a 10% - 15% performance increase over Ivy Bridge (again, assuming same clock speed), however I've read other people's comments about Haswell will be focusing more on reducing power consumption than on performance increase. That would probably mean Haswell's performance improvement would be half of my estimate (5% - 7.5%).

The truth is no one really knows how well Haswell will perform until they are benchmarked. Haswell will likely be released next year between April and June so you'll need to wait for some performance results.

The bottom line is to skip Ivy Bridge and wait for Haswell results to decide whether or not you should upgrade. Generally speaking, I do not upgrade (unless my CPU dies all of the sudden) until the performance difference between my current CPU (C2Q Q9450) and the future CPU is at least 20%.