• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Should I wait for the intel haswell processor to build a gaming rig?

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
There is a lot we don't know about our next change to microarchitecture yet but there is a lot of enthusiasm about it already so don't based your upgrade choice off from what might be but what you know. For me at home I am using a 1st generation Intel® Core™ i7; so I would most likely wait to upgrade my home system till the next microarchitecture. However if I was still using my old Intel Core 2 Duo I would be looking to upgrade right now.
 
I've already covered that:
1) A new generation of games will be released when the next consoles hit the street. Those consoles are likely to have AVX2 support or other powerful SIMD instuctions.
2) In any case your argument supports to not go with Ivy Bridge but wait longer!

... omitting for brevity
Well, the work week started so I have less time to spend on this, so I'll just briefly touch upon a few things, and drop quoting as it makes the posts gigantic:

We'll have to wait for details about consoles. Some of the PS4 rumours don't support this (e.g. the one with A8 3570 + Radeon 7670).
I don't think OP ever said what he has now, so it's hard to recommend. I generally agree with your last sentence, except that I don't see Haswell bringing a dramatic performance increase across the board either, so people with SB can probably skip it too, unless frequent upgrades are normal for them.


Search could be made parallel with threads, but it's <30 iterations, it's not worth the overhead.
Without c/p-ing the code and breaking NDA: imagine having a loop that goes from 1 to 30 and for i it retrieves object and calls some method (also has another boolean as param) and tests whether the return value is equal to some fixed value. I don't see what and how to pack into 256-bit registers and what instructions could be of any help when there's a situation like this. Outer loops, if there are any, are at least 5-10 level above in the call stack, it would just make things worse.
As for modifying container, each iteration must take into account what was done in the previous iteration because they operate on the same container and the order is important.

Maybe I don't completely understand how this works, but I can see how it helps with stuff like a += b*c etc. where you can pack 8 FPs/ints and do a single instruction on all of them, but in these general cases with deep stacks of function calls and larger objects, I just don't see it, which makes me believe that it will work great in fairly specific scenarios, like heavy integer/FP processing, all the instructions seem geared toward that too.

I agree that "there's more code that can benefit from AVX2 than code which benefits from GPGPU", but comparing raw number crunching ability, CPUs can't touch GPUs, and Haswell will not change much. It will come closer to current generation, but its throughput increase is something that's typical for GPUs over one-two generations because that is their bread and butter. When Haswell comes out, we will have Sea Islands and whatever comes after Kepler.

Regarding GCN etc., my point about GPU bound was that CPUs already have to wait for GPUs to render stuff, so no point in beefing up the CPU even more if GPUs can't keep up.

For neural networks: there are many kinds, but the most common ones don't have dynamic dependencies. Mostly you have layers and each node in the layer uses all nodes from the previous layers. But OK, I can see how some more general graph computing could benefit from this.


So unless you have an absolute ancient system and can't possibly wait any longer, it's clearly well worth waiting for Haswell.
As mentioned at the beginning, no argument there. But this is a very general rule for me, nothing specific to Haswell. If I had Q6600, I'd go with IB, with Q9550 probably not, and with i7 920 I'll probably skip Haswell as well (now say 'Haswell as well' aloud).
 
Well, the work week started so I have less time to spend on this, so I'll just briefly touch upon a few things, and drop quoting as it makes the posts gigantic:

We'll have to wait for details about consoles. Some of the PS4 rumours don't support this (e.g. the one with A8 3570 + Radeon 7670).
I don't think OP ever said what he has now, so it's hard to recommend. I generally agree with your last sentence, except that I don't see Haswell bringing a dramatic performance increase across the board either, so people with SB can probably skip it too, unless frequent upgrades are normal for them.


Search could be made parallel with threads, but it's <30 iterations, it's not worth the overhead.
Without c/p-ing the code and breaking NDA: imagine having a loop that goes from 1 to 30 and for i it retrieves object and calls some method (also has another boolean as param) and tests whether the return value is equal to some fixed value. I don't see what and how to pack into 256-bit registers and what instructions could be of any help when there's a situation like this. Outer loops, if there are any, are at least 5-10 level above in the call stack, it would just make things worse.
As for modifying container, each iteration must take into account what was done in the previous iteration because they operate on the same container and the order is important.

Maybe I don't completely understand how this works, but I can see how it helps with stuff like a += b*c etc. where you can pack 8 FPs/ints and do a single instruction on all of them, but in these general cases with deep stacks of function calls and larger objects, I just don't see it, which makes me believe that it will work great in fairly specific scenarios, like heavy integer/FP processing, all the instructions seem geared toward that too.

I agree that "there's more code that can benefit from AVX2 than code which benefits from GPGPU", but comparing raw number crunching ability, CPUs can't touch GPUs, and Haswell will not change much. It will come closer to current generation, but its throughput increase is something that's typical for GPUs over one-two generations because that is their bread and butter. When Haswell comes out, we will have Sea Islands and whatever comes after Kepler.

Regarding GCN etc., my point about GPU bound was that CPUs already have to wait for GPUs to render stuff, so no point in beefing up the CPU even more if GPUs can't keep up.

For neural networks: there are many kinds, but the most common ones don't have dynamic dependencies. Mostly you have layers and each node in the layer uses all nodes from the previous layers. But OK, I can see how some more general graph computing could benefit from this.


As mentioned at the beginning, no argument there. But this is a very general rule for me, nothing specific to Haswell. If I had Q6600, I'd go with IB, with Q9550 probably not, and with i7 920 I'll probably skip Haswell as well (now say 'Haswell as well' aloud).



What if you owned a Core2 duo E6850? What upgrade path would you pick?
 
2600k's are on sale at microcenter for $199 right now and there's a bundle with a gigabyte mobi as well. Personally, I'd get that since ivy will only be an upgrade for mobile processors.
 
What if you owned a Core2 duo E6850? What upgrade path would you pick?

The E6850 is clocked quite a bit higher but has only 2 cores vs the Q6600's 4. In the end it's kind of a wash and mostly depends what you run, but I'm guessing it falls into the "Go Ivy" camp. Still, if you don't have a pressing need to upgrade, you can't go wrong with waiting for Haswell.
 
I promise you that haswell will be as big a jump as q9650 to i5 750 or i7 860 or nearly as big at least. It will enhance the overall experience of the user in more than just benchmarks which would be just 10-20% better clock to clock. But the smoothness and snappiness will reach a different level. It is like comparing 3200 a64 Winchester to Venice except several times the difference. And people won't know about Winchester and Venice because they are old and not many people might remember using each in similar rigs but I have done that and it was a big difference and but benchmarks showed both with the same fps


That's a huge leap WOW. If that is the case ... World lets wait for Haswell unless you don't have a quad core cpu right now.
 
I wanted to build a gaming rig within the next 6 months but I have read that the intel haswell processor will be coming out in the spring of 2013. Now I figure that it would be worth it to wait to build my gaming rig until the haswell processor comes out. That way I would not have to upgrade my motherboard or other components if I decided to upgrade to a haswell processor in the future. So, as of right now, would it be worth it to wait for the haswell processor to come out and then build my gaming rig? Also, how much more powerful will the haswell processor be compared to the ivy bridge processors?



Its gonna be mainstream cpu anyway,

jD0it.jpg

http://www.custompcreview.com/2012/02/10/haswell-roadmap-leaked-launching/

i would rather buy x79 now with i7 3820 (OC 4.6ghz) and then upgrade to IB-e 8core (16threads) sometime 2014 🙂

qSk9F.jpg
 
i would rather buy x79 now with i7 3820 (OC 4.6ghz) and then upgrade to IB-e 8core (16threads) sometime 2014

I had the same line of thinking (except I wanted an IB-E in 2012). Now after reading all about Haswell, I will be moving to Haswell as soon as I can, and maybe bypassing IB-E.
 
I had the same line of thinking (except I wanted an IB-E in 2012). Now after reading all about Haswell, I will be moving to Haswell as soon as I can, and maybe bypassing IB-E.

Ib-e is suppose to be due sometime in 2013, but if i upgrade to this i7 3820 now then i would want it to last for a while, not upgrade next year ^^
 
Ib-e is suppose to be due sometime in 2013, but if i upgrade to this i7 3820 now then i would want it to last for a while, not upgrade next year ^^

I like my 3820. Runs cooler than a 2600K at same speeds. And with all the added features of SB-E, I will be happy until Haswell.
 
(Stares at q9550 equipped PC. Glances at Microcenter\2600K deal and posts in this thread)
"Why do you curse me by not giving me anything to complain about!! They say newer stuff is huge leap forward...why you no laggy laggy in my shootemups? Why do you still have OC headroom? Why do you still taunt me after all these years...asking only for the occasional graphics card upgrade?
Don't you know I could be winning the benchmarks?"
 
(Stares at q9550 equipped PC. Glances at Microcenter\2600K deal and posts in this thread)
"Why do you curse me by not giving me anything to complain about!! They say newer stuff is huge leap forward...why you no laggy laggy in my shootemups? Why do you still have OC headroom? Why do you still taunt me after all these years...asking only for the occasional graphics card upgrade?
Don't you know I could be winning the benchmarks?"

I'm in the exact same situation. It's been hard for me to accept I don't have to upgrade constantly to be a gamer anymore. I just upgraded my gpu (gtx280 to 7870) in my 2008 build. q9550 is at 3.4. Everything has been running great.

Now the problem I'm having is should I do things like upgrade my ram (4gb now to 8? ddr2 is expensive as hell!), will a SSD give me a worthwhile change? Adding up these small upgrades puts me close to a complete 2500k upgrade, so why not just go all the way.

I'm happy with my performance now though so I guess I'll wait till the new consoles force me to upgrade.
 
question is when can we expect mobile chips to start hitting? My MBP is 4 years old. It's not unusable, but I can tell its sluggish as hell. An IVB upgrade this year would be nice, but what if I stretched it out to Haswell?
 
question is when can we expect mobile chips to start hitting? My MBP is 4 years old. It's not unusable, but I can tell its sluggish as hell. An IVB upgrade this year would be nice, but what if I stretched it out to Haswell?
How much RAM do you have? My wife's 2008 MacBook became significantly smoother after upgrading from 2 GB to 8 GB. RAM is incredibly cheap now so it can help you wait a year longer, which will definitely be worth it.
 
I'm in the exact same situation. It's been hard for me to accept I don't have to upgrade constantly to be a gamer anymore. I just upgraded my gpu (gtx280 to 7870) in my 2008 build. q9550 is at 3.4. Everything has been running great.

Now the problem I'm having is should I do things like upgrade my ram (4gb now to 8? ddr2 is expensive as hell!), will a SSD give me a worthwhile change? Adding up these small upgrades puts me close to a complete 2500k upgrade, so why not just go all the way.

I'm happy with my performance now though so I guess I'll wait till the new consoles force me to upgrade.

I'm similar. Still rocking the e8400 o/c to 3.6. Occasionally get bottlenecked in games by the CPU but that's rare. The BEST upgrade I ever did was to a SSD. This will give you the most noticeable performance improvement you've ever seen in all your computing days 🙂
 
It depends on what you have right now. If you have SB then there is no reason to upgrade. I'm not even going to upgrade from PII. For what I do it works just fine.

What I will do is buy a nice 3TB HDD and a Samsung 830 SSD. Those things will be much more valuable to me then a CPU upgrade I won't even notice.
 
Last edited:
It depends on what you have right now. If you have SB then there is no reason to upgrade. I'm not even going to upgrade from PII. For what I do it works just fine.

What I will do is buy a nice 3TB HDD and a Samsung 830 SSD. Those things will be much more valuable to me then a CPU upgrade I won't even notice.

Damn bro you still gaming on a Pentium II 🙂
 
Back
Top