I5 2500k, 3570k or wait for Haswell?

[arcticle]

The question says it all. I do not want to dish out an extra $80-120 for an i7 and get very little performance improvements. That being said, do you think Haswell chips will also NOT be worth the price? I plan on using my computer for high end gaming and general use.

 

[jaideep1337]

yep get the ivy since you would get PCI-e 3.0 support and higher ram speed compatibility.

Ivy is also 6% faster than sandy but it does get hotter due to intels tri-gate 3d transistors

 

[InvalidError]

Ivy is also 6% faster than sandy but it does get hotter due to intels tri-gate 3d transistors

Ivy gets hotter due to higher power density (more watts per mm^2 of die area) and paste instead of solder between the die and IHS.

The tri-gate transistors and die shrink reduce power and heat output which would have otherwise made Ivy run cooler.

 

[jaguarskx]

Haswell is the 'tick' (new architecture) that is supposed to bring IPC and other architectural improvements. Broadwell is only a 'tock" (process shrink) which means it should not have any major improvements except maybe new IGPs and the expected integrated (MCP) IO hub.

Ivy Bridge is a 'tock+' (die shrink, new transistors, new IGP, other minor tweaks) and did not bring much in terms of extra clock speeds and IPC so there isn't much of a reason to expect a major jump (except for IGP) from Haswell to Broadwell.

Generally true about "tick" and "tock", but I think Intel is flipping it for Haswell and Broadwell. In terms of personal computing, the market is moving away from desktops and even laptops to the mobile device segment which includes tablets and smartphones. Sure Intel has Atom CPUs for tablets and they are dogs, not even worth putting into smartphones. Haswell will be focused on power consumption rather than performance so that they can be used in tablets. But I really doubt a Haswell CPU will ever make it into a smartphone, Intel needs to further reduce power to get into that market segment. They are working on low power consuming 48 core CPU for smartphones which was announced last week, but they do not think it will be ready for another 5 to 10 years.

The die shrink set for Broadwell will allow Intel to increase performance of their CPU and iGPU cores while maintaining or further reducing power consumption. I believe it is in Intel's best interest to be able to enter the tablet market segment will a good performing CPU since Apple and Samsung has a good leg up on Intel. Texas Instruments (TXN) decided exits out of the tablet and smartphone processor market (called SoC - System on Chip) because the amount of competition out there means that the SoC business will not generate as much revenue as TXN would like. Amazon is currently trying to strike a deal with TXN for the SoC business. It would further diversify Amazon, they will be able reduce the cost of the Kindles since those things uses SoC developed by TXN. Kindle's main competitor the Nook from Barnes & Nobles also uses TXN's SoC. It would be interesting to see what would happen if Amazon is able to buyout TXN's SoC division.

 

[InvalidError]

It would be interesting to see what would happen if Amazon is able to buyout TXN's SoC division.

Probably not much. Amazon would still have to fulfill existing contracts for current and near-future products which should give B&N plenty of time to source ARM chips from some other vendor for whatever comes after that.

 

[jaguarskx]

One of three things...

1. Honor TXN's contracts and not raise prices.
2. Raise prices.
3. Deny B&N future purchases of the SoC. Very unlikely since it is a revenue stream.