Share your AIDA 64 cache and memory benchmark here

[SiRoCu]

i7 8700K @5.3GHz

 

[dalekdukesboy]

i7 8700K @5.3GHz

That's exactly the voltage my 22nm 8 core e5 1680 v2 needs for up to 4.67 ghz or so lol.

 

[dalekdukesboy]

A new high of FSB with CPU megahertz! Highest I ever got stable was few posts ago 4.681 and only got it do it once or twice was hard to replicate. However fine tuned the VTT and VCCSA voltages and lowered them while still keeping everything stable so reduced heat/stress on system and allowed for another 10 mhz while still keeping the FSB at near my record high at 138.03. Highest I've gotten it to be stable is 138.5 or so. I try 139 or higher it won't even boot, it's a very fine line of what is stable versus what won't even boot into windows.

 

[ArbitraryAffection]

Here is my 2700X score. It is stock and RAM is on XMP profile, no other manual tweaks. I am pretty happy with the result. Zen+ has really good cache performance, rivaling and even beating Intel CPU's. L2 is of particular merit vs Skylake-Client, imo, because Zen+ has 2x the caapcity but similar bandwidth and latency. but.... much higher IMC latency. that is the killer for the lower than expected gaming performance on Ryzen IMO. But 66ns is pretty good for Ryzen i think^^

edit; i didn't realise aida64 added cache bandwidths up from all the cores. so actually skylake has much better bandwidth in L1 and L2 bandwidth but Zen+ has better L3 bandwidth. latency is about the same for all though

 

[chevy350]

Here's my 2700 with some generic Kingston 2400 running great at 3400 so far. Game and TM5 stable.

 

[ArbitraryAffection]

Tried some RAM ocing this evening. Even with the wonky bugged CL timing it has superior latency to the stock result. Pretty happy 63ns great for ryzen^^

 

[er557]

windows insider optimizations are getting better, also ran the bench at high priority, still more room for improvement, the only thing going for me here is octal channel and super fast L1/2/3 caches, being the motherboard is not overclockable and now only -50mv undervolt of cpu, cache, sa agent

 

[ArbitraryAffection]

windows insider optimizations are getting better, also ran the bench at high priority, still more room for improvement, the only thing going for me here is octal channel and super fast L1/2/3 caches, being the motherboard is not overclockable and now only -50mv undervolt of cpu, cache, sa agent

View attachment 118379

I think.. ehm.. that it doesn;t measure the cache speed of just one core, right? I mean like there is no way that L1 cache is 6.4TB/s bandwidth thats just impossible lol. is it cumulative or something? I always notice this cache bandwidth reader is really weird sometimes.

 

[agent_x007]

AIDA64 always adds L1/L2 cache speeds (if they are seperate for each core).

 

[ArbitraryAffection]

AIDA64 always adds L1/L2 cache speeds (if they are seperate for each core).

Thanks for clearing that up for me haha i always wondered this. So to get this straight: on my 2700X l1 cache speed is 1018GB/s / 8 = 127.25GB/s per core. So the L1 cache BW in reality for a single core is 127.25GB/s ? thanks

 

[agent_x007]

Basicly, but keep in mind this is synth. test.

 

[Mr_Headshot]

This is from my Asus 1080 Ti OC edition at factory overclock. Comments and questions welcome because I don't know if this is good or bad.

 

[er557]

It's really nice for a 1080 ti oc, mainly because it has nothing to do with this benchmark, which measures cpu/memory cache and bandwidth. To measure graphics card performance, you need to run something like 3dmark.

 

[Mr_Headshot]

It's really nice for a 1080 ti oc, mainly because it has nothing to do with this benchmark, which measures cpu/memory cache and bandwidth. To measure graphics card performance, you need to run something like 3dmark.

Thanks:

 

[juhasz221hun]

Hi everyone!

This is my result. Is it good enough?
What can i do to be better?
I had to set DRAM voltage to 1.400V to run 3200Mhz.

 

[Mr_Headshot]

It's really nice for a 1080 ti oc, mainly because it has nothing to do with this benchmark, which measures cpu/memory cache and bandwidth. To measure graphics card performance, you need to run something like 3dmark.

Thanks. I'm such a knucklehead at times. Lolol.

 

[TWK_OCZ]

 

[sneekypeet]

Ran it for giggles with PUBG running in the background....specs rig

 

[Deleted member 24505]

Meh

 

[kidza_ocz]

 

[biffzinker]

Meh

Try adjusting your timings for less latency if possible. Make a backup of your CMOS settings to a USB flash drive before changing anything.

 

[NoJuan999]

G.SKILL TridentZ RGB (2x8GB) F4-3200C16D-16GTZR @ 3400 MHz:

 

[Marktomato]

2x Kingston HyperX FURY [HX424C15FB2/8] 8 ГБ

 

[naxneri]

 

[Lindatje]