Userbenchmark, underpeforming ram?

demondrops

demondrops

Limp Gawd
Joined
Jul 7, 2016
Messages
422
I get this

Corsair Vengeance LPX DDR4 2666 C16 2x8GB-kr 1,640
12,098 User benchmarks - Average bench 78%
2 of 8 slots used
16GB DIMM DDR4 clocked @ 2133 MHz

Performing below potential (3rd percentile) - ensure that a dual+ channel XMP BIOS profile is enabled

i check in cpu-z it say dual channel, and i ran them stock there on XMP profile i get like 5th percentile and prety much same stuff. cant think there is something wrong? everything apear to be just fine. considering it apear that it test against identical ram kits then the result does seem extremely weird. anyone with an explanation?
 
so youre running 2666 at 213 and are surpriced that it performs worse? you need to compare 2133 to 2133 not 2666 or run it at 2666 then compare.
 
i was just testing out running them stock just to see what happen prety much lol :p so no not really but consider lower timings and the difference is ->

Corsair Vengeance LPX DDR4 2666 C16 2x8GB-kr 1,467
12,166 User benchmarks - Average bench 78%
2 of 8 slots used
16GB DIMM DDR4 2666 MHz clocked @ 2667 MHz

Performing below potential (6th percentile) - ensure that a dual+ channel XMP BIOS profile is enabled
48.2% Average
MC Read 18.2
MC Write 15.4
MC Mixed 14.3
46% 16 GB/s
SC Read 15.6
SC Write 14.3
SC Mixed 15.8
44% 15.2 GB/s
Latency 60.9
66% 60.9 ns
Poor: 47% ↑
This bench: 48.2%
 
Last edited:
looking at that link it looks normal to me. says its above average at ~78%. if its not erroring its working at normal.
 
hmm.. even ran memtest few months ago. so basically my ram is the worst as it can get lol. lost the ramicon lottery? do ram loose their funk like harddrives do? maybe they got like fresh kits or something.
 
no I think it working fine. ive never seen that sight before but it is saying your ram is performing above average (78%) compared to others with the same ram. mobo choice can also change scores.
 
x99 is a quad channel mb you only have two sticks of ram so dual channel...
the ram is performing as it should but by comparison to others with quad channel ram in a x99 system obviously its lacking
 
i think it even match amount of ram and sticks. as far as i could tell anyway. regret not getting that 4x4 but it costed a bit more.. so i guess it just have to be that then i thougt it wouldnt compare to anything but what i had in my system.
 
maybe your right and it is comparing against dual channel kits
x99 5820k was the first ddr4 memory controller and isn't as efficient as newer dual channel memory controllers
overclocking the uncore can help it from memory
there is also a lot of timings that can probably be tightened a bit
 
yy as it specifically tell me to checked if dual channel is active :D that i had no idea about, like u say maybe there is the difference it is used on newer cpu mobo ++ as it just check ram vs ram not the other components, and 5820k isnt popular i think compared to "Normal" chips from amd or intel.
 
if its not the quad channel thing it could also be comparing your 2666 to people that have higher like 3200. more people have faster ram dropping your 266 further down the list? the speeds youre getting seem normal to me.
 
tried to put them at 1.350 v but they wont budge one bit in terms of MHZ just not booting. i guess it is possible they did put a nice oc on their ram but they probably would get even higher latencies and 2133mhz 15c vs 2666mhz 16c was around 3% increase. and it did also seem like it would identify ram type correctly even on stock so idk. there probably is something wrong on their end. i guess i should not bother with it, i even did another MEMTEST this nigth just cuz i was bored, all good. maybe it is like dasa said about newer memory controllers, or just the interaction of cpu /w ram also? have no idea rly.
 
maybe run aida64 and see how that performs
3dmark timespy should let you know if your suffering any nasty performance drops vs other x58 systems
 
9516 on timespy regular.. aida64 it say 33x multiplier (wrong) memory type single channel, probably cuz it say TRIAL VERSION on my other ram stick. Im not quite sure what good AIDA64 is when most field say trial version tho xD CPU-Z report correctly multiplier and it say dual channel.
 

Attachments

  • cachemem.png
    cachemem.png
    92.3 KB · Views: 10
Last edited:
can you provide a link for timespy so i can see your cpu score and compare it with other x99 systems?

for me with aida64
6700k
2133c15 single channel 16533 MB/s read
2133c15 dual channel 31113 MB/s read
3866c17 DC 55369 MB/s

here is some results with x99 in dual and quad channel with 2400c13 ram which while faster than your 2666c16 it shouldn't be that much faster and it does kinda look like your system is performing as if its running single channel
https://uk.hardware.info/reviews/74...more-lanes-more-speed-benchmarks-applications
 
hm and i got this noctua fan that ofc cover entire ram slots on the side i put them :D i guess i will check my motherboard manual and inspect if they actually are in correct slots. need to dust my filters and all anyway.
 
wow im such a tool haha.. i guess cpu-z was wrong. not sure how long ive run them in single channel ... but performance have been great overall so idk. i dusted my filters and checked motherboard manual and swapped around to correct slots. what a dreadful monster this noctua is to have when having to change up something but its done now. TY btw iv'e would have run them in single all day long lol if u didnt tell me otherwise xD im going to have anoter run in 3dmark see if there is any difference in scores.

Corsair Vengeance LPX DDR4 2666 C16 2x8GB-kr 1,467
12,166 User benchmarks - Average bench 78%
2 of 8 slots used
16GB DIMM DDR4 2666 MHz clocked @ 2667 MHz

Performing above expectations (75th percentile)
 
3dmark shows minimal improvement from memory speed as its a small orderly program that fits most of what it needs into the cpu cache
if you hit any cpu limited parts of games then single channel could be dropping your frames by around 10-30%
 
You must log in or register to reply here.
Back
Top