Ivy E vs Haswell E for gaming

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

UaVaj

Golden Member
Nov 16, 2012
1,546
0
76
NEED
(1) only game that needs 6 core "today" is BF4. every other game needs 4 core (honestly ~2.5 core). on top of this. (2) only if you are planning to run 780ti/290x in multi gpu configuration.

if neither of these applies. stick with i5 haswell.
honestly. even an i5 sandy bridge will more than suffice.


WANT
haswell e all the way. bragging right are not cheap.
 

Kippa

Senior member
Dec 12, 2011
392
1
81
I am aware that today BF4 is and Planetary Annihilation are two of a handful of games that only use 6+ cores. What you have to bear in mind is that I got my new 6 core 4930k rig to last hopefully 5 years and as a guess there will be games in that time that will use 6 cores and possibly more, especially when new console ports start coming across.
 

Grooveriding

Diamond Member
Dec 25, 2008
9,107
1,260
126
Battlefield 4, Crysis 3.. both perform better on SB-E and IB-E than a 4670K or 4770K. That is a two gen old CPU outperforming the latest Intel has due to more cores and HT. Going forward games are being made primarily focused on new console platforms that have multiple cores.

At this point if you are upgrading today and in the future for gaming at a minimum you should be getting an i7 with hyperthreading. It is only going to get worse for i5s as more and more next-gen games are released. Heck the 4670K is slower than some AMD CPUs in BF4... Pretty sad considering how much better IPC is on Intel chips.

I'm not saying you should run out and get an 8 core extreme edition HW-E. But I wouldn't get anything less than an Intel i7 for gaming for future upgrades. i5 is starting to fall behind with 4 cores and no hyper threading.
 

OCGuy

Lifer
Jul 12, 2000
27,227
36
91
HT has been shown to be a wash up until this point. A 6 core SB/IB should no question beat a quad Haswell in anything that uses the 50% moar corez.

50% increase in cores > 20% IPC improvement.


What i think some people are missing is that someone who dropped 1k on a E-series strictly for gaming, isnt interested in debate about need.


Being able to tout the latest and greatest in your sig is a great feeling, yet fleeting. It is hard for some to return to the middle of the pack as long as they can afford to keep paying bleeding edge tax.

I really hate car analogies, but it is very similar to the rush of owning a new model year followed by the crash once it is no longer the latest.
 
Last edited:

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
Going forward games are being made primarily focused on new console platforms that have multiple cores.

True, but those are games are going to be using 6 weak cores (I think the other two are reserved for OS and other applications). So the benefit in moving to 6 or 8 much more powerful cores will likely be a case of diminishing returns. We shall see.

What i think some people are missing is that someone who dropped 1k on a E-series strictly for gaming, isnt interested in debate about need.

That's a great point. Many of us with 6+ cores running are doing other things - DC projects, video editing, software development, etc. The fact that some games are faster is just a bonus. Others simply like having the biggest baddest system around.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
Single-threaded performance will continue to be very important moving forward. A CPU with "aggressive Turbo" may not fare as well in a games that contain multiple threads with one normally being much more heavily utilized, because EIST will see core utilization and drop speeds too much, crippling ST performance of the critical process.

Of course, OCers eliminate that problem, but it will be interesting to see how non-K chips perform if they are aggressively Turboed as postulated.
 

Batmeat

Senior member
Feb 1, 2011
803
45
91
Totally depends on the game programming if it can utilize more cores or not. Single threaded games won't do squat with a multiple core processor, so processor speed is the limiting factor. We won't have a great idea until Haswell -E hits.
 

OCGuy

Lifer
Jul 12, 2000
27,227
36
91
Totally depends on the game programming if it can utilize more cores or not. Single threaded games won't do squat with a multiple core processor, so processor speed is the limiting factor. We won't have a great idea until Haswell -E hits.

Can't we look at how 8-core AMD CPUs perform on specific software, compared to the 4 and 6 core products in the same family, and make reasonable predictions?
 

Grooveriding

Diamond Member
Dec 25, 2008
9,107
1,260
126
HT has been shown to be a wash up until this point.

If a game is written to take advantage of more than four threads, as next gen games are starting to, HT makes a significant impact.

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Battlefield_4_China_Rising_-test-bf_4_proz.jpg


proz.jpg


m%20proz.jpg
 

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
If a game is written to take advantage of more than four threads, as next gen games are starting to, HT makes a significant impact.

*graphs*

The interesting bit to me is that Hyperthreading makes more of a difference than the Sandy Bridge -> Haswell improvements. Sandy Bridge with HT beats out Haswell without it, on a threaded engine.
 

Grooveriding

Diamond Member
Dec 25, 2008
9,107
1,260
126
The interesting bit to me is that Hyperthreading makes more of a difference than the Sandy Bridge -> Haswell improvements. Sandy Bridge with HT beats out Haswell without it, on a threaded engine.

Yeah. I wish more sites would spend time on it. Right now the games that make use of it are few and far between, but I see that changing as more engines are updated for next gen gaming. CPU benchmarks of Unreal Engine 4 will be significant to see. We already see FB3 needing over 4 threads for optimal performance.
 

OCGuy

Lifer
Jul 12, 2000
27,227
36
91
The interesting bit to me is that Hyperthreading makes more of a difference than the Sandy Bridge -> Haswell improvements. Sandy Bridge with HT beats out Haswell without it, on a threaded engine.

I'm old as of today so maybe I am missing it...but to me those charts are questionable.

The 2nd one shows a difference between 2500k and 2600k that can be explained by the 100mhz clock bump.

The 3rd one shows a 4C/8T beating a 6C/12T?

As for BF4 there is an article specifically on HT I will try to find it. It show that BF4 can take advantage of HT, but it ends up being inconsequential.

Different place shows it CPU bottlenecked at 1920X1200:

CPU_01.png


http://www.techspot.com/review/734-battlefield-4-benchmarks/page6.html
 
Last edited:
Nov 26, 2005
15,086
310
126
I don't think people are aware that if Core Parking is in effect there won't be proper use of all the cores. When I disabled Core Parking, Unreal Tournament III used ALL 4 cores on my i7 920 so to say that a handful of games use more than whatever amount of cores is not entirely accurate.
 

Grooveriding

Diamond Member
Dec 25, 2008
9,107
1,260
126

Single player benchmarks, different settings. Irrelevant.

What we still have right now is the same rhetoric we heard when the Q6600 was released and everyone thought their dual core e8400 was all they would need.

Cutting edge games and engines are starting to be threaded. This will only continue with the new console platforms.

Sadly the latest Haswell i5 is slower than AMD's otherwise anemic CPUs in the most cutting edge games because of these game's highly threaded
nature.

Probably not a huge deal if you are not concerned with leaving that 20fps on the table in BF4. But future upgrades of gaming machine's CPUs should be i7.
 
Last edited:

OCGuy

Lifer
Jul 12, 2000
27,227
36
91
Single player benchmarks, different settings. Irrelevant.

What we still have right now is the same rhetoric we heard when the Q6600 was released and everyone thought their dual core e8400 was all they would need.

Cutting edge games and engines are starting to be threaded. This will only continue with the new console platforms.

Sadly the latest Haswell i5 is slower than AMD's otherwise anemic CPUs in the most cutting edge games because of these game's highly threaded
nature.

Probably not a huge deal if you are not concerned with leaving that 20fps on the table in BF4. But future upgrades of gaming machine's CPUs should be i7.

Ummm...multiplayer benchmarks are irrelevant and you cannot reproduce them.

That website is terrible, 61 visitors per day...plus multiplayer tests which are ludicrous, and it shows in the benchmarks. How do you control how many players? What destruction is going on? Etc.

The Core i3 actually shows some impressive HT gains on the low end, the i7? Not so much.

In fact there are threads everywhere with people disabling HT on the 4770k because it is thought to cause more crashes then good.

Can you spot why multiplayer benchmarks are possibly inaccurate?

BF-4-1920-x-1080-Ultra-settings-GTX-770-vs-7970.jpg


http://www.hardwarepal.com/battlefield-4-benchmark-mp-cpu-gpu-w7-vs-w8-1/8/
 
Last edited:

Grooveriding

Diamond Member
Dec 25, 2008
9,107
1,260
126
That website is terrible, 61 visitors per day...plus multiplayer tests which are ludicrous, and it shows in the benchmarks. How do you control how many players? What destruction is going on? Etc.

Gamegpu.ru is an excellent site. They provide youtube videos of everything they test. http://gamegpu.ru/action-/-fps-/-tps/battlefield-4-china-rising-test-gpu.html

This is a running theme I see with your posts in reference to benchmarks; if you don''t like the results you attack the website with your biased opinion. Fortunately these website's tests and results hold weight, while your simple opinion and dismissal of them hold none.
 

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
Its also an ad hominen attack against the site. If the only argument is the site has low visitor counts I would say that is a logical fallacy. If we can genuinely poke holes in their methodology, prove their results are wildly different to all the other sites that have a consensus with similar hardware then fine but so far that hasn't been done (and ad hominen attacks are not allowed according to ATF rules FYI).

What you have to realise about gamegpu.ru is they use high end GPUs always, so its really mismatched with the CPU. For a long time they were using a GTX 690 and now they are using an Ares II. They always use SLI/crossfire and they regularly turn down the settings unrealistically on their CPU tests to make it show a difference. The CPU tests are sometimes excellent examples and sometimes they aren't. In the BF4 china rising test they are using a 780 ti with very high quality settings at 1080p, that is solid setup with reasonable settings and the graph should thus be reasonably representative. Hard to reproduce due to it being MP but I suspect if you get a similar player count game and the same spot your own results would be similar if not identical.
 
Aug 11, 2008
10,451
642
126
Single player benchmarks, different settings. Irrelevant.

What we still have right now is the same rhetoric we heard when the Q6600 was released and everyone thought their dual core e8400 was all they would need.

Cutting edge games and engines are starting to be threaded. This will only continue with the new console platforms.

Sadly the latest Haswell i5 is slower than AMD's otherwise anemic CPUs in the most cutting edge games because of these game's highly threaded
nature.

Probably not a huge deal if you are not concerned with leaving that 20fps on the table in BF4. But future upgrades of gaming machine's CPUs should be i7.

In what games is a 4670k slower than a stock 8350? From the graphs above I dont see that, and dont recall seeing it in any other benchmarks either. In the BF4 benchmark, the only one that includes haswell i5 shows it equal to the 8350 (within the margin of error) in both average and minimum framerate.

I will agree that six cores or hyperthreading will probably be beneficial, but I also think that unless one insists on the absolute cutting edge performance, a Haswell i5 will be more than adequate, especially if overclocked, for the forseable future.
 

Enigmoid

Platinum Member
Sep 27, 2012
2,907
31
91
HT has been shown to be a wash up until this point. A 6 core SB/IB should no question beat a quad Haswell in anything that uses the 50% moar corez.

HT shows off pretty well on i3's. I think its more that games don't really scale past 6 threads and scales poorly past 4.
 
Last edited:

TheRyuu

Diamond Member
Dec 3, 2005
5,479
14
81
In fact there are threads everywhere with people disabling HT on the 4770k because it is thought to cause more crashes then good.

"More harm than good" you mean?

Whenever I see stuff like "disable HT" those threads/posts/whatever often seem like they have the same feel as some of the bullshit you see about HPET and what have you.

Same goes for the core parking stuff although I can actually understand there being an underlying problem with how Windows handles the power states of the cores in that case (buggy).
 
Nov 26, 2005
15,086
310
126
Whenever I see stuff like "disable HT" those threads/posts/whatever often seem like they have the same feel as some of the bullshit you see about HPET and what have you.

HPET enabled is noticeable on my EVGA 760 A1 in Unreal Tournament III online DM where my ping is 20-30s. There is an odd system lag that isn't in sync with an online frag. I wouldn't say the same thing for HPET enabled on this Asrock X58 Extreme VS Disabled for playing Red Orchestra Rising Storm online. I can't feel a difference.
 

Kippa

Senior member
Dec 12, 2011
392
1
81
Can some one please explain what the benefits of hypthreading are? I know that a 2 core cpu an have 4 threads but surely it is just spreading the workload over 2 physical cores, how much benefit can be gained and why?