Do you expect the 8GB sticks to clock faster/tighter because they are single rank or because of the lower density? Considering the Crucial 32GB kit in the thread title, should we be striving for single or dual rank?
Crucial 32gb (2x16gb) cl16 3600 BL2K16G36C16U4B - Single Rank
- Thread starter DJ_Switch
- Start date
Do you expect the 8GB sticks to clock faster/tighter because they are single rank or because of the lower density? Considering the Crucial 32GB kit in the thread title, should we be striving for single or dual rank?
the test shows like for like same speeds and timings and focuses on just the difference in ranks. thats the correct way to do the test to take all other variables out of the picture and show the difference rank makes.
Both, lower density and single rank both positively influence the ability of a stick to improve both timings and speed.
The correct way to do a test is first to actually test something worth testing and second to compare apples to apples. All they have shown is that performance is higher with bank interleaving enabled - well of course it is. However in use bank interleaving comes with a penalty in terms of requiring lower timings and/or speed when compared to an equal quality single rank setup - what needed testing was whether the performance penalty from losses was worth taking thanks to the gains from interleaving. Dual rank kits have slower timings than single rank kits of the same quality (this is why they cost more per gb). By fixing the timings and ignoring this fact they are making their testing pointless, they are not comparing apples to apples. A single rank kit will see an advantage in timings over a dual rank kit when all other factors involved are the same, this is a fact. This almost always ends up in the single rank kit having a small performance advantage over a dual rank kit. Their test was a waste of their time and the time of anyone who read it because they forced a level playing field where one doesn't exist. Imagine reviews did that, that's the same as underclocking an Intel because boosting higher than Ryzen "isn't fair".
the only way to do a proper comparison is to apples to apples, that means both single and dual rank setups need to run same speeds and timings which is what they did. anything else is just bad testing. the review went to show the difference single vs dual rank had and it succeeded in doing just that.
if you feel you can run higher clocks on single bank ram then carry on but to say dual rank makes no difference is just wrong.
if you feel you can run higher clocks on single bank ram then carry on but to say dual rank makes no difference is just wrong.
Yes but that tells you nothing. Its like testing if 1T command rate is faster than 2T command rate or Gear down mode - well yes of course it is. The real question is does running slower speed at 1T overcome any gain in attainable mhz or timings from using GDM or 2T? I can't really make the point any clearer than that. They have just tested if a setting to improve memory performance (rank interleaving) that has worked since the days of SDRam works, and surprise surprise it does. They cannot say dual rank kits are faster than single rank kits because they will not be at the same timings when bought in the same budget range (similar £ per Gb). Their test was pointless beyond sanity checking that rank interleaving improves scores. They could have just turned off rank interleaving on the dual rank sticks and not tested the single rank sticks at all for all their testing is worth.
I feel like the most interesting test might be comparing Crucial BL2K16G36C16U4B.M16FE1 versus BL2K16G36C16U4B.M8FB1... I realise this is E vs B die, but this must be the closest apples-to-apples comparison possible to compare dual vs single rank.
I found this video really helpful for someone like myself who was a real noob when it comes to understanding the memory ranks etc:
https://youtu.be/kEP8iVVQR-g
https://youtu.be/kEP8iVVQR-g
Bought the Crucial Ballistix RGB BL2K16G32C16U4WL 3200 MHz, overclocked to 3600 no problem on the XMP profile and TimeSpy CPU score went from 10800~ to 12000~ going from 16gb 3000mhz to this kit.
I think its dual rank as its M16FE and shows die density to be 8gb e-die on thaiphoon burner, not sure what else im looking for. Also shows organization 2048m x64 (2 ranks). So I can recommend the 3200 kit if you want to save some cash.
I think its dual rank as its M16FE and shows die density to be 8gb e-die on thaiphoon burner, not sure what else im looking for. Also shows organization 2048m x64 (2 ranks). So I can recommend the 3200 kit if you want to save some cash.
Caporegime
Yeah thats dual rank.
Thanks, happy days
That's a tidy deal, good performance for the price!
Associate
(Copied from another thread)
I read my Crucial Ballistix Micron E-die 2x 16Gb 3600Mhz dual rank sticks with Thaiphoon burner, exported in nanoseconds the complete HTML report then imported this into the DRAM calculator. I can then click SAFE to get these settings which after a long time I managed to input into my MSI Tomahawk X570 motherboard BIOS. The settings seem stable on memtest with 1.4v memory and the recommended voltage settings on the others.
https://1drv.ms/u/s!AsMDJqyHDc8RguETIJibrQaDB8p-yQ
This seemed to only gain me 1% in 3d Mark timespy vs XMP which seemed less that I was expecting for dropping CL16 to CL14.
When I try to choose the Fast option it tells me it is not possible with Manual mode (which is what it switches to when importing the data from Thaiphoon burner). If I change to A0 memory quality it deletes all the left side readings but does give me new values with the FAST button. None of the FAST calculations (A0, A2...) include CL <16 though. Is this normal?
https://1drv.ms/u/s!AsMDJqyHDc8RguEW3J4j2RJyB7u0HA
Here is my MEMbench run with the SAFE timings which include CL14. My run time is a bit slower than the ones posted above but I think I can try to tighten one or two from what the calculator suggested (tRFC and tRC based on calculations suggested elsewhere). Also mine being a 5800X with 16 threads rather than 24 might make a difference. I had one time that Chrome froze for a few seconds despite finishing 4 runs of memtest without errors which I wonder could be a sign of overly tight timings. Lastly I am going to run Aida64 with XMP vs Tuned timings to see if there is a latency difference. The <60 result here seems quite good though.
I read my Crucial Ballistix Micron E-die 2x 16Gb 3600Mhz dual rank sticks with Thaiphoon burner, exported in nanoseconds the complete HTML report then imported this into the DRAM calculator. I can then click SAFE to get these settings which after a long time I managed to input into my MSI Tomahawk X570 motherboard BIOS. The settings seem stable on memtest with 1.4v memory and the recommended voltage settings on the others.
https://1drv.ms/u/s!AsMDJqyHDc8RguETIJibrQaDB8p-yQ
This seemed to only gain me 1% in 3d Mark timespy vs XMP which seemed less that I was expecting for dropping CL16 to CL14.
When I try to choose the Fast option it tells me it is not possible with Manual mode (which is what it switches to when importing the data from Thaiphoon burner). If I change to A0 memory quality it deletes all the left side readings but does give me new values with the FAST button. None of the FAST calculations (A0, A2...) include CL <16 though. Is this normal?
https://1drv.ms/u/s!AsMDJqyHDc8RguEW3J4j2RJyB7u0HA
Here is my MEMbench run with the SAFE timings which include CL14. My run time is a bit slower than the ones posted above but I think I can try to tighten one or two from what the calculator suggested (tRFC and tRC based on calculations suggested elsewhere). Also mine being a 5800X with 16 threads rather than 24 might make a difference. I had one time that Chrome froze for a few seconds despite finishing 4 runs of memtest without errors which I wonder could be a sign of overly tight timings. Lastly I am going to run Aida64 with XMP vs Tuned timings to see if there is a latency difference. The <60 result here seems quite good though.
Soldato
- Joined
- 10 Apr 2004
- Posts
- 13,496
Just received a 2x16GB kit of this stuff, SR M8FB1 Micron B-Die C9BLG.
Quick and dirty tests on my 9900K/Z370 Extreme 4 so far:
3600-C15-1.28V
3800-C16-1.31V
4000-C16-1.34V
Edit: Update, these are cracking sticks compared to the rubbish 3200C16/3600C18 stuff I've tried before.
4000, 16-18-18-38 with very tight sub timings at 1.35V.
Just running through another TestMem5 session and I'll move onto tRFC and tREFI (631 and 11430 currently).
Quick and dirty tests on my 9900K/Z370 Extreme 4 so far:
3600-C15-1.28V
3800-C16-1.31V
4000-C16-1.34V
Edit: Update, these are cracking sticks compared to the rubbish 3200C16/3600C18 stuff I've tried before.
4000, 16-18-18-38 with very tight sub timings at 1.35V.
Just running through another TestMem5 session and I'll move onto tRFC and tREFI (631 and 11430 currently).
Last edited:
Just got a 32GB (2x16GB) 3600Mhz CL16 kit - BL16G36C16U4B.M8FB1
Couldn't install them yet as i'm waiting for parts but had to ask Crucial if they were single or dual rank. They confirmed they sold both versions and mine were single rank.
Think i should return these and try for dual rank versions?
Couldn't install them yet as i'm waiting for parts but had to ask Crucial if they were single or dual rank. They confirmed they sold both versions and mine were single rank.
Think i should return these and try for dual rank versions?
Soldato
- Joined
- 10 Apr 2004
- Posts
- 13,496
Mine are the Single Rank too and are excellent overclockers [this is compared to other 32GB kits I've had that basically wouldn't move 100mhz irrespective of voltage!]. This kit is currently running tighter timings at 4000Mhz than the stock 3600 at the default 1.35V! Better timings than even the more expensive Crucial MAX kits!
Did an 16 hr Ram Test session with zero errors last night, and currently doing an overnight test with even tighter secondary/tertiary timings as well as a reduced tRFC of 610.
AIDA64 tests coming out at (average of 5 tests):
Read: 56,330 MB/sec
Write: 57,660 MB/sec
Copy: 52,768 MB/sec
Latency 39.9 ns
9900K, ASRock Z370 Extreme 4. VCCIO/SA: 1.13V
Think i might keep mine and overclock them then. These were cheaper than any other 3600Mhz CL16 memory i could find.
Only other option is go upto CL18.
I read the first sentence and thought "unlucky!", but you've hit gold with those sticks. Well done!
Soldato
- Joined
- 10 Apr 2004
- Posts
- 13,496
All things being equal dual rank would give you “free” bandwidth.
But in reality, it puts more strain on the IMC and may not overclock so well.
And even then, your not going to notice a few % in bandwidth above 50GB/sec in day-to-day.
Soldato
- Joined
- 10 Apr 2004
- Posts
- 13,496
Right, this is where I am after a few days tweaking. I've been having reboot memory training issues so I've nudged the IOL offsets up one from the Auto setting (21 --> 22) to see if it stabilises it (VCCIO/SA/Ram voltage seemingly does nothing to resolve the situation...!)
Ram Voltage 1.35V
VCCIO/SA 1.15V *Although 24hrs+ MemTest86, RamTEST, etc stable reboots/sleep were a problem. 1.15V resolved this.
9900K, P0 stepping
ASRock Extreme 4, P4.20 BIOS
-40mV Vcore offset
52/52/51/51/50/50/49/48 CPU multi (0 AVX offset).
46x Cache
Going to double tREFI and do a bunch of overnight testing and if that passes I think I'll leave it there.
Out of curiosity for you Ryzen folk, I did a quick and dirty boot tests (without changing the secondary/tertiaries from the 4000 CL16 settings which may have hampered it a little...):
3800, C15-18-18-38, 1.375V
3800, C14-18-18-38, 1.400V
3800, C14-17-17-38, 1.425V
This kit may even do 4000, C15 but I'm happy staying around the 1.35V mark.
Either way for £140, I'm more than happy!
Ram Voltage 1.35V
VCCIO/SA 1.15V *Although 24hrs+ MemTest86, RamTEST, etc stable reboots/sleep were a problem. 1.15V resolved this.
9900K, P0 stepping
ASRock Extreme 4, P4.20 BIOS
-40mV Vcore offset
52/52/51/51/50/50/49/48 CPU multi (0 AVX offset).
46x Cache
Going to double tREFI and do a bunch of overnight testing and if that passes I think I'll leave it there.
Out of curiosity for you Ryzen folk, I did a quick and dirty boot tests (without changing the secondary/tertiaries from the 4000 CL16 settings which may have hampered it a little...):
3800, C15-18-18-38, 1.375V
3800, C14-18-18-38, 1.400V
3800, C14-17-17-38, 1.425V
This kit may even do 4000, C15 but I'm happy staying around the 1.35V mark.
Either way for £140, I'm more than happy!
Last edited: