Haswell-e around the corner, upgrades from ivy-e, haswell, anyone taking the plunge?

[theskyisthelimit 48]

It appears that Haswell-e is just around the corner (next month, NDA released as of Aug 29th for reviews).  In looking at various sources it doesn't sound as though its that much of an upgrade at least for FSX/Prepar3D users, though possibly a slight performance boost if comparing i7-4930k and the newer i7-5930k (but lower clock speeds with haswell-e).

 

There doesn't appear to be any change in pcie lanes, still 40.. so if you are a multi-gpu person (3) then no benefits (for other gaming).  The ram is going to be DDR4 (2133), which probably wont be a huge advantage anytime soon.

 

The more "powerful" i7-5960x is to be 8/16 core with a base clock of 3.0ghz and 3.5 in turbo, while the i7-5930k will be 6/12 with a base of 3.5ghz and 3.7ghz turbo.  I've read the starting retail price on the i7-5960x is likely to be $1100.. yikes, while the i7-5930k will be in the $630 range (compare with around $350 for a new haswell currently or even $579 for a new i7-4930k ivy-e).

 

In my case i have tested both haswell and ivy-e (had both) and opted to stick with ivy-e which has been doing well for me, even for prepar3d (with a 780ti).. i run with HT enabled in my case as well.  My 4930k is a 6/12 with base of 3.4ghz/3.9ghz turbo and 12mb cache.. the newer haswell-e will have a bigger cache to boost performance slightly.

 

I've ran into some problems with sata ports on my board in recent days, so frankly i was considering rma'ng and selling my equipment to move onto haswell-e, but the lower clock speeds has me considering otherwise (at least for flight simming).  I run my cpu at 4.4ghz up from the 3.4ghz.. i'd probably target the i7-5930k.. i believe that cpu is 140watt, while the 4930k is 130watt, so maybe overclocking wont be as easy.

 

CPUMark from Passmark software on some cpus right now are (most of these i personally tested except the 4790):

????? - i7-5960x 3.0ghz (8 physical/16 ht)

????? - i7-5930k 3.5ghz (6/12)

16,000 - i7-4930 4.4ghz overclocked (6/12)

14,469 -i7-4930k 3.4ghz with turbo (6/12)

11,896 - i7-4770k 4.4ghz overclocked (4/8)

11,383 - i7-4790 4.0 ghz (4/8)

10,154 - i7-4770k 3.4ghz (4/8)

10,485 - i7-2600k 4.6ghz overclocked (4/8)

 8,645 - i7-2600k 3.4ghz (4/8)

 

 

Anyone else been considering a move to haswell-e in the coming months?  Seems to be an iffy upgrade for simming at least.

[theskyisthelimit 48]

Cpumark scores have been coming in.. can now fill in the blanks on the new cpu's.. doesnt look like its worth going from a 4930k to either of this at the moment at least based on cpu horsepower.

 

CPUMark from Passmark software on some cpus right now are (most of these i personally tested except the 4790):

15970 - i7-5960x 3.0ghz (8 physical/16 ht)

13378 - i7-5930k 3.5ghz (6/12)

*16,000 - i7-4930 4.4ghz overclocked (6/12)

*14,469 -i7-4930k 3.4ghz with turbo at 3.9 (6/12)

11,896 - i7-4770k 4.4ghz overclocked (4/8)

11,383 - i7-4790 4.0 ghz (4/8)

10,154 - i7-4770k 3.4ghz (4/8)

10,485 - i7-2600k 4.6ghz overclocked (4/8)

 8,645 - i7-2600k 3.4ghz (4/8)

[TheGrunt 44]

Passmark is IMO quite bad CPU test mainly for two reasons: 1) it is very synthetic and close to the architecture, which means that it scales for example pretty much linearilly with the amount of physical cores and represents very badly the performance of many real life software. 2) It is known to give if not bad and erroneus, then at least very odd results.

 

I myself have a 3930K running 4.6GHz and based on the reviews and different benchmarks I see no reason to upgrade at least because of P3D. I may opt the 5960x some day if I happen to have a chance to get it for a "decent" price like I had my 3930K back in the day, but I definitely not do it for P3D.

[theskyisthelimit 48]

Passmark is IMO quite bad CPU test mainly for two reasons: 1) it is very synthetic and close to the architecture, which means that it scales for example pretty much linearilly with the amount of physical cores and represents very badly the performance of many real life software. 2) It is known to give if not bad and erroneus, then at least very odd results.

 

I myself have a 3930K running 4.6GHz and based on the reviews and different benchmarks I see no reason to upgrade at least because of P3D. I may opt the 5960x some day if I happen to have a chance to get it for a "decent" price like I had my 3930K back in the day, but I definitely not do it for P3D.

 

I agree synthetic benchmarks on the CPU are not enough to go on.. i guess if the architecture had more to offer in terms of performance/video etc it would be worth a closer look, but not for my needs at this point.  I'd love to even not overclock and get the same performance (gain that whopping extra 5-7 fps lol) without the hassle though.