I think that this reserves a thread. Best RAM on Haswell-E

S

sblantipodi

2[H]4U
Joined
Aug 29, 2010
Messages
3,765
Is there any reviews that focus on ram ?
I would like to know what is the best buy for 5930k.
2100MHz, 2400MHz, 3000MHz or what?

I preferably don't like to over clock the bus so if I can go down to 2100mhz I will be glad.
 
I see you've decided to buy despite all your complaining :D

[H] already did all this hard work for you, running their Haswell E test setup with stock 1866 DDR4 (Haswell E on the graphs), and then overclocked to 2800 DDR4.

http://www.hardocp.com/article/2014/08/29/intel_haswelle_core_i75960x_cpu_x99_chipset/3#.VAHj2YEcSmU

In the real-world benchmarks results they swap places, neither one having a solid win, and the difference is never more than 1%.

So it looks like stock memory will be just fine. Overclocked DDR4 will be overkill until they can get the timings down.
 
defaultluser said:
I see you've decided to buy despite all your complaining :D

[H] already did all this hard work for you, running their Haswell E test setup with stock 1866 DDR4 (Haswell E on the graphs), and then overclocked to 2800 DDR4.

http://www.hardocp.com/article/2014/08/29/intel_haswelle_core_i75960x_cpu_x99_chipset/3#.VAHj2YEcSmU

In the real-world benchmarks results they swap places, neither one having a solid win, and the difference is never more than 1%.

So it looks like stock memory will be just fine. Overclocked DDR4 will be overkill until they can get the timings down.
Click to expand...

I threat hardware like girls threat men.
I complain about things I like :D

I would like to see real world comparison about different ram on the same platform, not different ram on different platforms.
I think this is the best way to choose kits.
 
sblantipodi said:
I threat hardware like girls threat men.
I complain about things I like :D

I would like to see real world comparison about different ram on the same platform, not different ram on different platforms.
I think this is the best way to choose kits.
Click to expand...

Just pick the ones that appeal to you, close your eyes and run your mouse over them. Pick the one that your mouse lands on.

Or wait until there are alot of reviews.
 
sblantipodi said:
I would like to see real world comparison about different ram on the same platform
Click to expand...

I swear you're acting like a stereotypical girl. I just showed you exactly what you asked for, and you didn't listen.

HARDOCP compared Haswell E 8 core with 1866 MHz ram versus the same processor with 2800MHz ram.

SAME PROCESSOR, FASTER AND SLOWER RAM!

REAL GAME BENCHMARKS ON LOW TO REMOVE ANY GPU LIMITING

REAL MULTIMEDIA TESTS

SAME PLATFORM, DIFFERENT RAM SPEEDS

Am I making myself clear? I don't know how I could possibly make this any clearer to you.
 
defaultluser said:
I swear you're acting like a stereotypical girl. I just showed you exactly what you asked for, and you didn't listen.

HARDOCP compared Haswell E 8 core with 1866 MHz ram versus the same processor with 2800MHz ram.

SAME PROCESSOR, FASTER AND SLOWER RAM!

REAL GAME BENCHMARKS ON LOW TO REMOVE ANY GPU LIMITING

REAL MULTIMEDIA TESTS

SAME PLATFORM, DIFFERENT RAM SPEEDS

Am I making myself clear? I don't know how I could possibly make this any clearer to you.
Click to expand...

In the link you are giving me I see an Haswell-E and an Haswell-E at 4.45GHz, how this results can be useful for what I am asking about?

A 5960X at stock clock and a 5960X at 4.45GHz are two completely different CPU in terms of performance, so what is the sense of this comparison?
 
sblantipodi said:
In the link you are giving me I see an Haswell-E and an Haswell-E at 4.45GHz, how this results can be useful for what I am asking about?

A 5960X at stock clock and a 5960X at 4.45GHz are two completely different CPU in terms of performance, so what is the sense of this comparison?
Click to expand...

Okay, let me spell this out for you:

1. All the processors in the review are overclocked to 4.5 GHz in order to compare instructions per clock. That Haswell E, along with everything else, run at 4.5 GHz. You can confirm this in the test setup:

All systems are tested with identical component and OS setup with obvious hardware exceptions noted above. Our AMD system was tested at 5GHz while the rest of our systems were tested at 4.5GHz.
Click to expand...

2. The bus overclock required to run the ram at 2800 MHz means the processor had to have it's bus overclocked as well. This meant the closest thing to 4.5 GHz the author could acheive was 4.45 GHz. That's close enough to call them the same.

Currently on the ASUS X99-Deluxe motherboard we are using required us to change the bus speed and strap setting to get 2800MHz DDR4 RAM speeds. This of course gave us a non-4.5GHz score for comparison. We matched the CPU clock rate as close as we could for comparison, so we end up with a 4450MHz clock rate with the DDR4 at 2800MHz. We have been talking to ASUS about this and we should be seeing DDR4 RAM able to be clocked at 2400MHz rates and up without impacting the bus speeds in upcoming BIOS releases extremely soon
Click to expand...

3. This is a fair comparison at 4.5 GHz, because nobody in the right mind would spend a thousand dollars on a processor with no intention of opverclock. Also, this means the absolute worst-case for needing to feed the processor, so it's the best way to test the effects of faster ram.

Next time, may I strongly suggest you read the whole review before jumping to conclusions?
 
Last edited:
defaultluser said:
Okay, let me spell this out for you:

1. All the processors in the review are overclocked to 4.5 GHz in order to compare instructions per clock.

2. The bus overclock required to run the ram at 2800 MHz means the processor had to have it's bus overclocked as well. This meant the closest thing to 4.5 GHz the author could acheive was 4.45 GHz. That's close enough to call them the same.

3. This is a fair comparison at 4.5 GHz, because nobody in the right mind would spend a thousand dollars on a processor with no intention of opverclock. Also, this means the absolute worst-case for needing to feed the processor, so it's the best way to test the effects of faster ram.
Click to expand...

ok I have understood now, but I prefer an article like this :
http://www.anandtech.com/show/7364/memory-scaling-on-haswell/10
 
So I am in the same boat. I realize that yes performance wise you won't see a lot of difference but does 2800 make a difference vs 2400 when you are overclocking?
 
It depends on each individual, if the differences that showed up in synthetic benchmark figures bothers you a lot, then you should get the fastest DDR4 you can afford. After all you're spending a lot of money here, might as well ensure you're fully satisfied with your new system.

If you're only concern about real world performance, then I have yet to see any evidence that suggest spending on the fastest DDR4 available will bring any noticeable improvements. Of course if you run a benchmark that specifically test for memory bandwidth, then you're going to see different figure. But does that actually translate to any difference in actual applications that we use every day such as gaming or media editing? I don't think so.

Personally I would prefer spending the money elsewhere that have actual impact, SSD, graphic cards, etc.
 
Hornet said:
It depends on each individual, if the differences that showed up in synthetic benchmark figures bothers you a lot, then you should get the fastest DDR4 you can afford. After all you're spending a lot of money here, might as well ensure you're fully satisfied with your new system.

If you're only concern about real world performance, then I have yet to see any evidence that suggest spending on the fastest DDR4 available will bring any noticeable improvements. Of course if you run a benchmark that specifically test for memory bandwidth, then you're going to see different figure. But does that actually translate to any difference in actual applications that we use every day such as gaming or media editing? I don't think so.

Personally I would prefer spending the money elsewhere that have actual impact, SSD, graphic cards, etc.
Click to expand...

quote it, mature analysis.
 
Save money. Don't get expensive RAM right now. Faster kits (4ghz+) will be out in a year or two. Use the money you saved at launch to buy a faster kit later. :)

I personally went with a 2400mhz g.skill kit.
 
You must log in or register to reply here.
Back
Top