Jump to content


Photo

Build Time: Firebreather


Best Answer Dasa , 08 March 2018 - 09:07 AM

for gaming 8600k will spank that cpu or go 8700k if you also want to do some rendering or something with video

more cores higher ipc and lower price

http://translate.goo...rt75579-27.html

 

Corsair H60 120mm Liquid

this would be hot and loud

a big hsf like nh-d15 would be much better

for aio to compete on a performance per db ratio you need to spend over $200

 

as for ram i would suggest this as 3600 15-15-15 is faster than 4000 19-23-23

https://www.newegg.c...N82E16820232306

 

or if you wanted to run dual rank ram and maybe a ram cache

this 3200c14 kit should oc fairly well if you spend some time giving it the love and vdimm it needs

https://www.newegg.c...N82E16820232378

 

as for the best tim to use ontop of the ihs

GELID Solutions GC-Extreme
or
Thermal Grizzly Kryonaut

and if you wish to delid the cpu for a 5GHz+ oc this is what you would use under the ihs

Thermal Grizzly Conductonaut

 

if budget is no problem go with a 1tb 960 evo or intel 760 or 2tb and get rid of the old spinner altogether

optane wont provide a noticeable benefit to these ssd so you can drop it

 

maybe look at a liquid cooled 1080ti

Gigabyte AORUS GeForce GTX 1080 Ti Waterforce Xtreme Edition <just make sure its the aio version unless you want to build a custom loop

 

oh and as you would need a z370 mb to go with those cpu

ASUS Z370 ROG Maximus X Apex $479 will probably get you the highest memory oc

but ASRock Z370-EXTREME4 is exceptional value

 

as for a monitor upgrade you may want to wait for these hdr10 panels

https://www.anandtec...de-curved-200hz

Go to the full post


  • Please log in to reply
53 replies to this topic

#41 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 24 May 2018 - 07:37 PM

Got a few somewhat underwhelming pics I can upload, but I've been trying to conquer a hardware problem so haven't got to it (or the benching) yet: the mouse doesn't work.  Pointer moves, and can left/right click on the desktop, but start menu and any window (including those opened from desktop right-click) don't even know there is such a thing as a mouse, let alone that one is trying to click on something.

 

Been quite a while since I've had to navigate Windows using only a keyboard.  Jesu does it suck.

 

 

--- Later ---

 

Sorted.  Bought a new mouse - just a 5-button cheapie from Big-W, but it works well enough.  If it breaks, it ain't like I lost a lot.

 

On with the work! :)


Edited by Cybes, 24 May 2018 - 10:51 PM.

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#42 Rybags

Rybags

    Immortal

  • Super Hero
  • 35,756 posts

Posted 24 May 2018 - 11:06 PM

Annoyingly some modern mouses need the driver/support software to function properly.  Like Logitech, for whatever reason they decided to change default button codes and you get jibber-jabber operations without the software running.



#43 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 24 May 2018 - 11:13 PM

These are somewhat larger than I imagined, sorry.
 
Power off, left side:
riw4lQo.jpg
 
Power off, right side:
FlSYbsq.jpg
 
Powered, through window:
HY3auv3.jpg
 
Powered, overview:
ewXjVcK.jpg

Rather untidy loop hanging between the power connectors on the GPU, but considering he had to work with the stock cables, I think he did pretty well. On with downloading benches.

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#44 Rybags

Rybags

    Immortal

  • Super Hero
  • 35,756 posts

Posted 24 May 2018 - 11:21 PM

Nice that they provide a void under the motherboard next to the PSU - my case has the rubberized grommets over medium sized through-holes that make routing the cable out the back somewhat harder and also steal from the run length.



#45 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 25 May 2018 - 01:42 AM

Results just in from the 3d Mark tests (Fire Strike and Time Spy):

Results without overclocking of CPU, GPU, or RAM; or tweaking on nVidia control panel.


Fire Strike (Asrock Default Mode) 20,615
-----------
Graphics Score: 27,788
Graphics Test 1: 131.77 FPS
Graphics Test 2: 111.55 FPS

Physics Score: 13,086
Physics Test: 41.54 FPS

Combined Score: 9,945
Combined Test: 42.26 FPS

Online comparison: Better than 96% of all results
(4k gaming system comes in at 17,805!)


Fire Strike (Asrock Performance Mode) 20,712
-----------
Graphics Score: 28,090
Graphics Test 1: 135.42 FPS
Graphics Test 2: 111.22 FPS

Physics Score: 13,052
Physics Test: 41.44 FPS

Combined Score: 9,912
Combined Test: 46.10 FPS



Time Spy (Asrock Default Mode) 8,372
--------
Graphics Score: 9,215
Graphics Test 1: 58.65 FPS
Graphics Test 2: 53.98 FPS

CPU score: 5,514
CPU test: 18.53 FPS

Online comparison: Better than 88% of all results
(4k gaming system comes in at 6,733)


Time Spy (Asrock Performance Mode) 8,611
--------

Graphics Score: 9,564
Graphics Test 1: 61.55 FPS
Graphics Test 2: 55.46 FPS

CPU Score: 5,514
CPU Test: 18.49 FPS


Note that there are some folks on the "Hall of Fame" results box which are *much* higher. I don't know how much they spent, but it would have to be jaw-dropping.


---


@Rybags: Yeah, it's a very competent case - excellently designed for access and airflow. Not so much for aesthetics. (That's kinda how I want anyway, so I'm happy.)

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#46 gamble

gamble

    Primarch

  • Atomican
  • 1,373 posts

Posted 25 May 2018 - 01:50 AM

Looks great, like that case!!  Im just now use to my monitor going from 24 to 32 and the curve.  I have my old monitor on the side with Afterburner and CPUID HWmonitor running.  IDK if it is old habit but always run monitors and like to check on it all the time.  I like to tweak the fans and stuff to the lowest before there is diminishing returns on cooling(unless benching!).. Gaming is going to be so much better for you for years to come!

 

Nice bench results, looks about right.  Curious if you could run Afterburner when benching or gaming to get an idea of what your cards core mhz is running at.  My Asus runs near max out of the box at around ~1950mhz but I can only OC it about 30-40 more..  Curious as to what yours is at as you might have a lot more to push it up (not that a 1080ti really needs it.. :) 


HAF 912 | Biostar x370GT7 | 1700x @ 4ghz | Asus Strix GTX 1080ti @ 1962c/12000m | EVGA RAM @ 2933mhz  | EK Watercooled CPU | EVGA G2 1300W 


#47 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 25 May 2018 - 03:16 AM

Results for the Unigine benches (Superposition and Valley) without overclocking of CPU, GPU, or RAM; or tweaking on nVidia control panel.



Superposition (4K Optimized) (Default Mode) 9,012
-------------
FPS: Min 52.93, Max 89.18, Avg 67.41
GPU C: Min 30.0, Max 81.0
GPU Utilization: Max 99%
(Better than 15th place in online comparison)

Superposition (4K Optimized) (Performance Mode) 9,020
-------------
FPS: Min 52.95, Max 87.38, Avg 67.47
GPU C: Min 29.0, Max 80.0
GPU Utilization: Max 100%
(The ambient temperature here is dropping)

Very little difference between operating modes, here, unlike the 3d Mark benches; in those, there was noticeable if minor stuttering in scene with consistent motion (long pans, etc) in Default mode, but buttery smoothness in Performance. Here, almost nothing.


Valley (Default Mode) 5,806
------
Render: Direct3D11
Mode: 1920x1080 8xAA fullscreen
Preset: Extreme HD
FPS: Min 44.8, Max 243.5, Avg 138.8


Valley (Performance Mode) 5,769
------
Render: Direct3D11
Mode: 1920x1080 8xAA fullscreen
Preset: Extreme HD
FPS: Min 44.5, Max 245.8, Avg 137.9

Again, almost no differece between modes on this one. Maybe the Asrock tweaks are tuned for the benchmark?

I was curious the difference extra pixels would make; this next is the exact same except for resolution.

Valley (Performance Mode) 3,919
------
Render: Direct3D11
Mode: 2560x1440 8xAA fullscreen
Preset: Custom
Quality: Ultra
FPS: Min 38.2, Max 189.8, Avg 93.7

This looks awesome, btw - very smooth, with just one rapid pan across large draw distance which has two tiny hitches.



@Gamble:
The Asrock mobo comes with a package called (imaginatively) "A-tuning", which can handle some overclocks and fan tuning, as well as provide system info. I haven't looked yet at Afterburner, but I imagine it's similar.
As it happens, this last bench, Valley, displays the numbers you're after as it runs. During this bench:
GPU: 1,936 MHz
RAM: 5,505 MHz
GPU temp starts off at around 40C and climbs at roughly 1C/s to about 65C, then slows in a nice decelerating curve to top out at 80C several seconds before the 'mark finishes.

I would guess that leaves me quite a margin for OC. I'm somewhat shy to do so, however - I've fried a much less expensive card before.

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#48 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 25 May 2018 - 04:10 AM

And the boring desktop drone ones too - wPrime, SuperPi, yCruncher.
 
From memory, yCrucher - just get a single and multi-core score from benches 3, 4 and 5.  Omit 5 if you want, 3 should take seconds, 4 probably under a minute.


Super Pi MOD 1.9
----------------
16K: 0.125s (That does not even register on the score page, being under 1s)
32M digits: 8m 09s



W-Prime
-------
32M: 8.404s
1024M: 263.063s (4m 23.063s)



Y-Cruncher
----------

Option 3:
---------
Constant: Pi
Algorithm: Chudnovsky Formula
Decimal Digits: 50,000,000
Hexadecimal Digits: 41,524,102
Computation Mode: Ram Only
Threading Mode: Push Pool -> 12 / ? (randomization on)
Working Memory: 284,228,656 ( 271 MiB)
Total Memory: 367,553,584 ( 351 MiB)

Start Date: Fri May 25 03:52:24 2018
End Date: Fri May 25 03:52:28 2018

Total Computation Time: 2.799 seconds
Start-to-End Wall Time: 3.601 seconds

CPU Utilization: 543.11 % + 4.47 % kernel overhead
Multi-core Efficiency: 90.52 % + 0.74 % kernel overhead

Spot Check: Good through 50,000,000
Timer Sanity Check: Passed


Option 4:
---------
Threads: 6 (pinned to cores)

Terms Needed: 1,000,000,048
Internal Precision: 48 Hexadecimal Digits

Start Time: Fri May 25 04:00:05 2018

Frac( 1/(8k + 1) * 2^(-4k + 4000000002) )
Time: 1.720 seconds ( 0.029 minutes )

Frac( 1/(2k + 1) * 2^(-4k + 3999999999) )
Time: 1.751 seconds ( 0.029 minutes )

Frac( 1/(8k + 5) * 2^(-4k + 4000000000) )
Time: 1.750 seconds ( 0.029 minutes )

Frac( 1/(4k + 3) * 2^(-4k + 3999999999) )
Time: 1.777 seconds ( 0.030 minutes )

End Time: Fri May 25 04:00:12 2018

Total Time: 7.002 seconds ( 0.117 minutes )

CPU Utilization: 593.39 % + 0.45 % kernel overhead
Multi-core Efficiency: 98.90 % + 0.07 % kernel overhead


Option 5 is a character counter, for which you specify a file. Are you sure that's the one you want? If so, name the file, please.

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#49 gamble

gamble

    Primarch

  • Atomican
  • 1,373 posts

Posted 25 May 2018 - 05:36 AM

Thx for the numbers!  For the 1080ti the boost clock for founders edition is around 1600mhz... Yours running around 1936mhz, which is great, would probably be close to the limits of the core (most don't go much over 2025mhz anyway unless under water).  Unless you can keep the temps lower than 80c, which would probably need more aggressive fan curve to maintain, might not be worth it for the noise factor and the slight boost...  I run that speed instead of trying to push it since, like you its is pretty good OC already.  If you OC or not Id try to keep under 80c with fan curve using Afterburner or other software but really you have a great OC from the box :)  


HAF 912 | Biostar x370GT7 | 1700x @ 4ghz | Asus Strix GTX 1080ti @ 1962c/12000m | EVGA RAM @ 2933mhz  | EK Watercooled CPU | EVGA G2 1300W 


#50 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 25 May 2018 - 08:04 AM

Yours running around 1936mhz, which is great, would probably be close to the limits of the core (most don't go much over 2025mhz anyway unless under water).


I hadn't realised it was running on the edge out of the box. Good to know.

Unless you can keep the temps lower than 80c, which would probably need more aggressive fan curve to maintain, might not be worth it for the noise factor and the slight boost... I run that speed instead of trying to push it since, like you its is pretty good OC already.


Done deal - I'm more than happy with the performance as it is, and the fact that it doesn't make this small room blazingly hot AND is quieter than my laptop even at idle, I don't feel much inclined to mess with it. Maybe after a year and stuff starts to take a toll on it - and the warranty's expired. ;)

Just tried out Skyrim SE and Empyrion maxed out... Both are smooth as glass. Skyrim doesn't want to behave itself at 3440 (loses inputs), but what can you expect from a game that old? Osiris: New Dawn up next.

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#51 Rybags

Rybags

    Immortal

  • Super Hero
  • 35,756 posts

Posted 25 May 2018 - 11:00 AM

With SuperPi the one to run is the 1M test.  We have threads around here with a big database of SuperPi as well as the major release 3DMarks.

Not 100% sure but I think you need to set yCruncher to Run as Administrator.

I've just done back to back tests on yCruncher current version vs an older (0.7.1) and there was 5 seconds difference in the (0,0,3) test.

With yCruncher, I'm just referring to the Pi based tests - so select option 0 straight up - for clarity I just put the option numbers is brackets e.g. (0,0,3)

With wPrime - you need to set the # of threads.  It'll usually just default to # of cores but with HTT enabled you can just run double that # of threads for a better result.  I usually do runs with 4 and 8 threads.

 

For comparison - my machine:  Haswell i7-4970 (4.0 GHz turbo), 24 Gig Ram @ 1600 [11,11,11,28,2T]

These are without any optimisation attempts, background stuff like browser, torrents still running, no special tuning of Ram.

 

SuperPi mod 1.5, 16K:  0.138 secs, 1M:  9.76 secs

 

ycruncher 0.7.5.9481 tests:  (0,0,3 - Pi, single, 100M) 36.005 seconds wall time

(0,1,3 - Pi, multi, 100M) 9.418

(0,1,4 - Pi, multi, 250M) 26.159

 

wPrime 32 [8 threads] 7.253 secs

[4 threads] 10.768


Edited by Rybags, 25 May 2018 - 11:03 AM.


#52 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 25 May 2018 - 11:13 AM

With SuperPi the one to run is the 1M test.  We have threads around here with a big database of SuperPi as well as the major release 3DMarks.
Not 100% sure but I think you need to set yCruncher to Run as Administrator.


The latter part of this post I understood, but this quoted bit..? Yer talkin' Greek, man. ALL of those 3 needed to be run as admin or they just flat wouldn't run.

Anyway, I've been up all night getting this thing to where I can halfway use it - got a lot more still to do, but m' brayne aint werkin so hot right now. ;) I'll get to this after some sleep.

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#53 Cybes

Cybes

    Titan

  • Atomican
  • 18,337 posts
  • Location:Where I am

Posted 27 May 2018 - 08:54 AM

SuperPi mod 1.5, 16K:  0.138 secs, 1M:  9.76 secs


SuperPi 1.9, 1M 9.016s

ycruncher 0.7.5.9481 tests: (0,0,3 - Pi, single, 100M) 36.005 seconds wall time
(0,1,3 - Pi, multi, 100M) 9.418
(0,1,4 - Pi, multi, 250M) 26.159


0.7.5.9481:
(0,0,3) 31.829s wall time
(0,1,3) 6.522s
(0,1,4) 17.616s

wPrime 32 [8 threads] 7.253 secs
[4 threads] 10.768


32M 4-Thread: 8.507s
32M 8-Thread: 6.446s

Caveat: There was a YouTube video running at 480p throughout all of these. I doubt it made a huge amount of difference, but at least it was consistent difference. ;p

Sorry for the extra day - just...stuff.

"Reality does not care what you think." - Dr Richard Feynman
"There is no "I" in team." - "True.  I will not be found in any team."


#54 Rybags

Rybags

    Immortal

  • Super Hero
  • 35,756 posts

Posted 27 May 2018 - 12:06 PM

Never realised they bothered to update SuperPi further.  Has the annoying hardware scan delay but you can bypass with Shift.

 

Doesn't run properly on my system for some reason - it's only showing seconds without fractions and not generating any output files.  And that's also when running as Administrator.

 

I'd have thought your scores would have been better than mine by a bit more.  6 Cores vs 4, and much more generous turbo multipliers.  Plus each generation usually has some reasonable FP improvements.

Likely your Ram timings aren't optimised real well.  SuperPi and wPrime are old benches but still sensitive to memory settings and good to compare effects while you're fiddling with them.






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users