Question Performance on two builds is this expected


Feb 20, 2014
Hi there,

I have a query about two machine's i own. Just to provide some background, I purchased rig one back in 2015 from scan computers pre-built. At the time I was travelling quite a lot back and forth with to my gfs and wanted a machine at hers rather than transport my machine which was a pricey system at the time. I had an old acer machine which had the same PSU as rig two and an r9 270, and I had never put a machine together before so thought I would give it a shot and invested in new board, ram, cpu etc over the years and now have another rig.

Please see the specs for both rigs below:

Rig one (pre-built)

Asus X99-A/USB3.1 motherboard
i7-5820k OC 4.2 @ 1.344V
16GB DDR4 RAM Quad Channel
256GB M2 SDD
Gigabyte RTX 2080 Super
RMi750W GOLD certified

I've recently purchased the RTX 2080 Super and I am pretty pleased with the performance on higher resolutions such as 2k / 4k is awesome in comparison to my 980.

Rig two

Asus Z170-K Motherboard
i5-6600k OC 4.6 @ 1.35V
8GB DDR4 RAM Dual Channel
128GB M2 SDD
Corsair CX750

I read that in terms of 1080p res, I shouldn't see much of a difference as the 980 is still capable of a lot of games at 1080 at high/ultra settings.

But when I bench marked a few games, e.g. DOOM and Batman Arkham Knight, I noticed that the fps was higher on the Rig two, particularly on Doom.

To provide this, I tested the same arcade mode zone which was just the first one level (think it's called 'The UAC') and in some areas I was getting up to 200 FPS on rig two. I thought it might be that DOOM is more CPU bound and therefore benefiting from the higher clock speed on rig two, or maybe not utilising all the cores, on the 5820k.

I just wondered is this expected behaviour based on the specs above?

The cpu cooling on the 5820k (rig one) is Pure Rock CPU Cooler (same as rig two actually), would it be worth trying to increase the overclock and invest in better cooling?

I never really considered switching the cards round and mainly gaming on rig two, but is that actually the better system. Would be a shame if it was - Rig one burnt a big hole in my pocket.


Feb 20, 2014
Hi, I've just run CPU-Z on both machines:

rig one:

Type: DDR4
Channel: Quad
Size: 16GB
NB Frequency: 2920.2 MHz
DRAM Frequency: 1396.6MHz
FSB RAM 1:22
CAS# Latency (CL): 16.0 clocks
RAS# to CAS# Delay (tRCD): 18 clocks
RAS# Precharge (tRP): 18 clocks
Cycle Tme (tRAS): 36 clocks
Row Refresh Cycle Time (tRFC): 365 clocks
Command Rate (CR): 2T

rig two:

Type: DDR4
Channel: Dual
Size: 8GB
NB Frequency: 3899.1 MHz
DRAM Frequency: 1500.0MHz
FSB RAM 1:30
CAS# Latency (CL): 15.0 clocks
RAS# to CAS# Delay (tRCD): 17 clocks
RAS# Precharge (tRP): 17 clocks
Cycle Tme (tRAS): 35 clocks
Row Refresh Cycle Time (tRFC): 390 clocks
Command Rate (CR): 2T

So is it combination of this and the higher clock speed?


Feb 20, 2014

so i sent a copy of each rig by exporting a report from CPUID and sent to a friend and one he noticed was the NB frequency on the x99 rig was 2920 which he thought was low the overclock is 4.2?

Just wondering if anyone agreed with this?

I decided to change it so it is now set to 3936 - he then said to try testing your system make sure no stability issues from changing that.

it seems like the X99 machine is performing better than it was in gaming scenarios.

I ran a cinebench test and it passed - and then thought i should run a prime95 test. Just to remind everyone i bought this machine pre-built and the company OC it for me - setting the Voltage at 1.35 and OC at 4.2 - When they sent it to me they gave me some files proving they had tested the overclock using Adia64 etc. But not prime95.

Anyway, I tried testing the system with prime95 (as per friend's recommendation) and my temperatures were extremely high - like 100 degrees - so i killed the test.

I then spent quite a lot of time last night creating a new OC profile and trying to downclock the Vcore as it did seem high and i wondered if maybe i could achieve a more stable OC with low Vcore to bring the temps down a bit.

as I did this prime95 did perform better but after 5 - 10 mintues i was still getting into the 90s - so i decided to try and run this test at stock - and same thing happened - high 90 temperatures.

My friend is saying that it's strange my cooler is 130TDP - see here:!-pure-rock-compact-cpu-cooler-4-heatpipe-120mm-silent-pwm-fan-for-intel-amd-cpus-130w-tdp
yet, my CPU TDP is 140 i believe.

So I have a couple of questions:

  1. Is the NB frequency change - likely to make much of a difference or is this in my head?
  2. Is the cooler not really feasible for an i7-5820k? that's for both stock / OC. If not what would you recommend I purchase?
  3. Could this be that I need to just reapply some thermal paste?
  4. I've read quite bit online today about prime95 not being suitable for haswell chips - is this still the case? Should I not be using prime95 to test stability for this chip?