• Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!

S4 MINI Classic (S4M-C)

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
one option

https://enoch.systems/en/innodisk/8gb-288_pin_ddr

potential pick of that bunch....

https://enoch.systems/en/p/components/memory/m4c0-8gs1lcik.html

Can't find timings anywhere though. Samsung die so might overclock to match your corsair kit.
I dont think they'll overclock to match 4000 MHz lol, I'll probably leave them stock. Thx for the link

nice one Cuban :-)
i saw on your flickr album that you still have the hdplex 300?
you glued it also with epoxy? did it help with the coilwhine at all?
i did mine and it helped.
Thanks, and yea I still have the old HDPLEX 300, I did glue it with Epoxy, it's in one of my old posts in my build thread in my forum signature, it totally DID reduce coil whine. :)

Thanks @chx How exactly do i make sure I don't buy registered RAM?
 
  • Like
Reactions: Biowarejak

brt02

Cable-Tie Ninja
Jan 3, 2018
224
234
This is correct. I recently bought 2x8GB Innodisk VLP 2666MHz (M4U0-8GS1VCIK). Chips on the PCBs indicate these have K4A8G085WC-BCTD. I'm feeling buyer's remorse because this C-die doesn't seem to be anything like B-die. My overclocking attempts resulted in failure, every B-die can do 3200MHz CL14 but these don't even boot at 2666MHz CL14. 3200MHz 21-21-21-38 was Cinebench stable but that's a terrible amount of latency for Ryzen.

I'm using AMD Ryzen 2600 and Asus ROG Strix B450-i. Ryzen isn't the best platform for memory overclocking but I don't think that's the issue here. At the moment it seems C-die is a massive letdown. Of course it's possible that there's a compatibility problem, hopefully someone will test if overclocking works better with Intel or different AMD chipset.

Here's a screenshot from Thaiphoon Burner:




Timings are pretty slack given the low memory clocks. Interested to see the real world performance implications
 

brt02

Cable-Tie Ninja
Jan 3, 2018
224
234
I dont think they'll overclock to match 4000 MHz lol, I'll probably leave them stock. Thx for the link

Feed it some volts. Most XMP profiles demand 1.35V and a lot of people are fine with pushing 1.45V through Samsung dies for 24/7 operation.

Thanks @chx How exactly do i make sure I don't buy registered RAM?

Look for unbuffered RAM in the title / DIMM type. If it says registered then don't buy it as it won't work. The link that @chx posted above only has unbuffered RAM
 
  • Like
Reactions: CubanLegend

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
Feed it some volts. Most XMP profiles demand 1.35V and a lot of people are fine with pushing 1.45V through Samsung dies for 24/7 operation.



Look for unbuffered RAM in the title / DIMM type. If it says registered then don't buy it as it won't work. The link that @chx posted above only has unbuffered RAM
sweet thanks, I'll grab a pair of 2666 2x8GB sticks, does Single-Sided Single-Rank or Double-sided Dual-Rank matter?
 

Seid Dark

Chassis Packer
Oct 15, 2018
16
18
Timings are pretty slack given the low memory clocks. Interested to see the real world performance implications
After a lot of trial and error (had to clear CMOS multiple times) I managed to overclock Innodisk VLP to 3200MHz. There's a substantial performance improvement. Intel platform would most likely allow higher clocks and/or tighter timings.

Here's an AIDA64 benchmark, stock and overclocked:

 

brt02

Cable-Tie Ninja
Jan 3, 2018
224
234
I was watching the livestream haha - have only just seen your post here.

I would just go for the cheapest option. there are small differences. Dual rank might give you better performance, single rank might be easier to overclock.
 

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
I was watching the livestream haha - have only just seen your post here.

I would just go for the cheapest option. there are small differences. Dual rank might give you better performance, single rank might be easier to overclock.

Thanks for the info guys, I may attempt overclocking the RAM later, after I get it and am done running my CPU heatsink thermal tests on the BlackRidge.. if you guys have any recommendation for any information on overclocking these Innodisk VLP sticks, please share!

I ordered 2 of these 8GB sticks, should be here in 4-5 weeks, and now, to sit and wait!:


Subtotal $248.00
Shipping & Handling $9.54
Grand Total $257.54

If the BlackRidge & NFA12x15 beats my NHL9i and my current LP-56 & NFA9x14... then i'll say this was worth the price! Lowering temps in an S4M-C is like a drug... XD
 
Last edited:

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
After a lot of trial and error (had to clear CMOS multiple times) I managed to overclock Innodisk VLP to 3200MHz. There's a substantial performance improvement. Intel platform would most likely allow higher clocks and/or tighter timings.

Here's an AIDA64 benchmark, stock and overclocked:

I just ran this test, my 4000MHz DDR4 (19-23-23-45 CR2) with my 7700k @ 4500MHz, got me:
Read: 49644 MB/s
Write: 56790 MB/s
Copy: 46705 MB/s
Latency: 43.3 ns

Do you think i could get close to that if I overclock my upcoming 2666 VLP? I've heard that anything past 3200MHz is usually a negligible performance increase in games, DigitalFoundry did a video about memory frequency and I think there was only a handful of games sensitive to frequency.
 
  • Like
Reactions: br3nd0

Seid Dark

Chassis Packer
Oct 15, 2018
16
18
I just ran this test, my 4000MHz DDR4 (19-23-23-45 CR2) with my 7700k @ 4500MHz, got me:
Read: 49644 MB/s
Write: 56790 MB/s
Copy: 46705 MB/s
Latency: 43.3 ns

Do you think i could get close to that if I overclock my upcoming 2666 VLP? I've heard that anything past 3200MHz is usually a negligible performance increase in games, DigitalFoundry did a video about memory frequency and I think there was only a handful of games sensitive to frequency.
I'm 99% sure you will get better results than me, Z270 is great for RAM overclocking. Your latency will be quite a bit lower than my results, even with the same 3200MHz CL16 settings. High latency is the main issue with AMD Ryzen atm.

On Hardwareluxx forums some have achieved similar results to your current setup: https://www.hardwareluxx.de/communi...cking-ergebnisse-im-startbeitrag-1198323.html
 

palace97

Trash Compacter
Sep 6, 2017
40
16
Any1 know when the MSI RTX 2070 mini and the Gigabyte RTX 2070 itx are set to release? I understand the Zotac 2070 mini is released but it won't fit and has no idle fan control so after the other two. Also anyone know if the two fit inside the S4 Mini classic? currently got the MSI aero 1070 itx and that fits perfect with no mods.
 

threestripevida

Airflow Optimizer
Mar 28, 2017
277
466
Any1 know when the MSI RTX 2070 mini and the Gigabyte RTX 2070 itx are set to release? I understand the Zotac 2070 mini is released but it won't fit and has no idle fan control so after the other two. Also anyone know if the two fit inside the S4 Mini classic? currently got the MSI aero 1070 itx and that fits perfect with no mods.

I think we all just have to wait and see. I didn’t know that they announced any other mini cards for the new 2000 series. Isn’t the power draw on these cards higher? And it might not be enough for the HDPlex 400w?
 

palace97

Trash Compacter
Sep 6, 2017
40
16
I think we all just have to wait and see. I didn’t know that they announced any other mini cards for the new 2000 series. Isn’t the power draw on these cards higher? And it might not be enough for the HDPlex 400w?

HDPLEX 400 and the 330w dell power brick should be capable of powering the new 2070 mini's, i'm using the (non k variant of the i7 7700 CPU) So my power draw isn't too high anyways
 

threestripevida

Airflow Optimizer
Mar 28, 2017
277
466
2070 should be in the same ballpark as a 1080, if nothing else powerlimit it.

That’s good to know. I’m using a 1080 right now. I wouldn’t be looking to upgrade until next generation though. Hopefully companies continue to make powerful mini versions of GPUs.
 

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
Thanks for the info guys, I may attempt overclocking the RAM later, after I get it and am done running my CPU heatsink thermal tests on the BlackRidge.. if you guys have any recommendation for any information on overclocking these Innodisk VLP sticks, please share!

I ordered 2 of these 8GB sticks, should be here in 4-5 weeks, and now, to sit and wait!:


Subtotal $248.00
Shipping & Handling $9.54
Grand Total $257.54

If the BlackRidge & NFA12x15 beats my NHL9i and my current LP-56 & NFA9x14... then i'll say this was worth the price! Lowering temps in an S4M-C is like a drug... XD
The VLP DDR4 just came in...!

that's 3 weeks from ordering. :) Not bad. I'll be waiting on my custom S4M-C CPU side vent plate before doing my thermal tests. I'm going to do all thermal tests of the VLP DDR4, Blackridge cooler + NF A12x15 before the CPU side vent, and then post-CPU side vent. And all versus my current "Thermolab LP53 & NF-A9x14-PWM" setup. Inside the S4M-C. :D
 

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
So... my Blackridge arrived and I jumped the gun on my thermal tests. I redid all the top LP53 CPU tests I had done before (my temps went up by 2c from having done them last year as my LP53 was a little dusty and probably needing a thermal paste reapplication). Comparison of the two coolers:


BUT, installing this thing was a PAIN, I had instlaled the VLP DDR4, I attached my NF-A12x15 to it, and after a lot of trouble, and a couple of minutes of ACTUALY fearing that it wouldn't fit.. I was able to fit the BlackRidge into my S4M-C, but... in only ONE orientation:


Note that the fan was literally sitting against the PCI-E Riser and has to be pushed away from it so the cooler would seat properly, also it was brushing against the motherboards Rear I/O, and it was almost completely covering the VLP RAM, and jumpers on the motherboard below it, making resetting the CMOS jumper a HUGE PAIN. AND on top of this, the fins were perpendicular to the RAM, where in previous cases with other coolers, they've performed better with the heatsink fins parallel to the RAM. Also, note this cooler is 47mm high, which means it will fit inside the S4M-C, BUT the fins will be absolutely flush against the side panel. For pictures of all the above fitment issues and fitment tolerances in more camera angles, please see my S4M-C album on Flickr here.

But long story short when it came to temperatutes: the BlackRidge was defective on 4-5 of the 6 heatpipes, as can be seen here in this video demonstration of the heatpipe solder snafu that DAN showed us how to investigate for:

..and so, sadly, the cooler dissapointed me.

Test Setup:
-S4M-C with the Side panels ON
-in vertical orientation (power button on bottom)
-room temps were constant 74F (23c)
-temps taken after 10 minutes from hottest core temp unless otherwise stated

TEST__LP53 & NFA9x14 pull vs Blackridge & NF A12x15 push
Idle at Desktop 35c vs 38c
Prime95 26.6 80 vs 82c
RealBench Stress 79c vs 81c
x264 Custom Load 78c vs 82c
Intel Burn Test (Standard) Hottest core after 10 runs 67c vs 70c
AIDA64 System Stress Test 81c vs 84c

Notice, these temps are decent for a defective cooler, as my NHL9i performed about 2-4 c hotter than my LP53, so the defective BlackRidge is basically performing like a more expensive NHL9i, because i had to pay for VLP DDR4, and the NF-A12x15.

TL,DR: If you already have an NHL9i or LP53 in your S4M-C the manufacturer defective Blackridge is,.. well.. not worth it, right now. If the non-defective revision comes out later and can turn out to actually OUTPERFORM the LP53 when i test it again, then and only then will I say a purchase is justified.
 
Last edited:

Reldey

Master of Cramming
Feb 14, 2017
387
405
So... my Blackridge arrived and I jumped the gun on my thermal tests. I redid all the top LP53 CPU tests I had done before (my temps went up by 2c from having last year as my LP53 was a little dusty and probably needing a thermal paste reapplication). Comparison of the 2:
IMG_4833 by CubanLegend, on Flickr

BUT, installing this thing was a PAIN, I had instlaled the VLP DDR4, I attached my NF-A12x15 to it, and after a lot of trouble I was able to fit the BlackRidge into my S4M-C in only ONE orientation:
IMG_4881 by CubanLegend, on Flickr
Note that the fan was literally brushing the PCI-E Riser and has to be pushed away from it so the cooler would seat properly, also it was brushing against the motherboards Rear I/O, and it was almost completely covering the VLP RAM and jumpers on the motherboard below it, making resetting the CMOS jumper a HUGE PAIN. AND on top of this, the fins were perpenticular to the RAM where in previous cases with other coolers, they've performed better with the heatsink fins parallel to the RAM. For pictures of all the above, please see my S4M-C album on Flickr here.

But long story short: the BlackRidge was defective on 4-5 of the 6 heatpipes.. and so it dissapointed me.

Test Setup:
-S4M-C with the Side panels ON
-in vertical orientation (power button on bottom)
-room temps were constant 74F (23c)
-temps taken after 10 minutes from hottest core temp unless otherwise stated

TEST__LP53 & NFA9x14 pull vs Blackridge & NF A12x15 push
Idle at Desktop 35c vs 38c
Prime95 26.6 80 vs 82c
RealBench Stress 79c vs 81c
x264 Custom Load 78c vs 82c
Intel Burn Test (Standard) Hottest core after 10 runs 67c vs 70c
AIDA64 System Stress Test 81c vs 84c
Wow, thanks for going through all the trouble for this. Are you getting a replacement? Will it be easy to return? Being stateside makes me nervous to order this thing until I see some actual improvement over the NH-L9I cooler.
 
  • Like
Reactions: CubanLegend

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
Wow, thanks for going through all the trouble for this. Are you getting a replacement? Will it be easy to return? Being stateside makes me nervous to order this thing until I see some actual improvement over the NH-L9I cooler.
I am still going through the refund process with OC-UK, they've asked for proof that I've destroyed the cooler beyond repair and disposed of it. After that, they will be issuing me a refund, this is easier and faster for international customers, so we don't have to ship back units and have to pay return shipping.

Once the revision is out and is proven to not be defective, I'd say youre safe to buy from them. It should be in stock late January Early Feb from what @dondan predicted.
 
Last edited:

robbee

King of Cable Management
n3rdware
Bronze Supporter
Sep 24, 2016
883
1,388
@CubanLegend People are reporting better tempratures with a pull setup on the blackridge. You might be blowing hot air through the radiator with the current setup. Have you tried flipping the fan?
 
  • Like
Reactions: CubanLegend

CubanLegend

Steely-Eyed NVFlash Man
Dec 23, 2016
834
1,011
smallformfactor.net
@CubanLegend People are reporting better tempratures with a pull setup on the blackridge. You might be blowing hot air through the radiator with the current setup. Have you tried flipping the fan?
I thought I had tested the superior mode (which i thought was push), boy am I glad you suggested I try pull, Here's the new results (good news everyone)!


Test Setup:
-S4M-C #193 with the Side panels ON, in vertical orientation (power button on bottom)
-CPU is i7-7700k @ 4.5GHz, 1.2vcore
-Room temps @ 74F (23c)
-Temps taken after 10 minutes from hottest core temp unless otherwise stated

LP53 & NFA9x14 pull vs Blackridge & NF A12x15 push vs pull
Idle at Desktop 35c vs 38c vs 32c
Prime95 26.6 80 vs 82c vs 75c
RealBench Stress 79c vs 81c vs 74c
x264 Custom Load 78c vs 82c vs 74c
Intel Burn Test (Standard) after 10 runs 67c vs 70c vs 63c
AIDA64 System Stress Test 81c vs 84c vs 78c

Thats a 6-8c decrease in temps in BlackRidge push vs pull
And a 3-5c decrease in temp in LP53 vs BlackRidge pull
(When it comes to noise, the LP53 wins EASY as that NF-A9x14 is silent at idle.. but this BlackRidge has wind noise even with the NF-A12x15, since it's fins are packed tightly up against the S4M-C side panel)

But... are 3-5c of lower temps worth the price? Only if you don't care about your expendable income... You do the math: $60 Cooler, $250 VLP DDR4, $20 Noctua fan... thats $330 USD for 3-5c, thats an average of $88 USD per each degrees C.

When the non-defective version of this cooler arrives in February, then I'll re-purchase, and test and see if it's any better, but so far, I'm impressed.
 
Last edited: