Motherboard recomendations

[wingman99]

thats an amazing video test.
i wish i knew it all before buying my mobo,
and as he states in this footage gigabyte z370 hd3 has horrible issues with VRM temperatures and throttling, with LLC .. well with keeping rith voltages omg so much moaning this night
well what do u think is it worth to try to attach some small radiators on MOSFETS and generally on VRM section? does it sound silly tho ?

At 4.5GHz I would not worry about it. He is running prime95 at 5.0GHz when do you run that setup?

 

[Brando]

finally broke down and ordered some samsung b die on sale for $206. went with the 3200mhz cas14 in grey and black. they look like they match my mobo and should oc well. those cas15 3600mhz look nice but the price is too much. no biggie!
https://www.newegg.com/Product/Product.aspx?Item=N82E16820232409

 

[wingman99]

finally broke down and ordered some samsung b die on sale for $206. went with the 3200mhz cas14 in grey and black. they look like they match my mobo and should oc well. those cas15 3600mhz look nice but the price is too much. no biggie!
https://www.newegg.com/Product/Product.aspx?Item=N82E16820232409

Post back how they overclock.

 

[Woomack]

Can also check how high can you set memory clock and compare what is better. ASRock Z370 Extreme4 supposed to support 4333+. I left my memory @4500 after last tests and it runs like that for 3 weeks now

Btw. all new memory series are now on Micron and Hynix. I wouldn't be surprised if good Samsung IC disappear from the market or will be only in not many available 4000+ kits. In EU stores I can see only couple of 4000+ kits available. Almost only TridentZ and some overpriced Corsair.

 

[wingman99]

Can also check how high can you set memory clock and compare what is better. ASRock Z370 Extreme4 supposed to support 4333+. I left my memory @4500 after last tests and it runs like that for 3 weeks now

Btw. all new memory series are now on Micron and Hynix. I wouldn't be surprised if good Samsung IC disappear from the market or will be only in not many available 4000+ kits. In EU stores I can see only couple of 4000+ kits available. Almost only TridentZ and some overpriced Corsair.

You would think that the new Micron and Hynix would do better than Samsing ICs, why are they not doing better for overclocking? Why are the good Samsung ICs disappearing?

 

[Brando]

Just got it in. Looks like it is what it's supposed to be. Not sure what to try first. I'm not nuts with frequencies like some. What are the odds of getting 3400mhz at cas 13 with 1.4v or less?

Memory Module
Manufacturer
G.Skill
Series
Trident Z Silver & Black
Part Number
F4-3200C14-8GTZSK
Serial Number
00000000h
JEDEC DIMM Label
8GB 1Rx8 PC4-2133-UA1-11
Architecture
DDR4 SDRAM UDIMM
Speed Grade
DDR4-2133
Capacity
8 GB (8 components)
Organization
1024M x64 (1 rank)
Register Model
N/A
Manufacturing Date
Undefined
Manufacturing Location
Taipei, Taiwan
Revision / Raw Card
0000h / A1 (8 layers)
DRAM Components
Manufacturer
Samsung
Part Number
K4A8G085WB-BCPB
Package
Standard Monolithic 78-ball FBGA
Die Density / Count
8 Gb B-die (20 nm) / 1 die
Composition
1024M x8 (64M x8 x 16 banks)
Clock Frequency
1067 MHz (0.938 ns)
Minimum Timing Delays
15-15-15-36-50
Read Latencies Supported
16T, 15T, 14T, 13T, 12T, 11T, 10T
Supply Voltage
1.20 V
XMP Certified
1600 MHz / 14-14-14-34-48 / 1.35 V
XMP Extreme
Not programmed
SPD Revision
1.1 / September 2015
XMP Revision
2.0 / December 2013
Thermal Sensor
Manufacturer
Seiko Instruments
Model
S-34T04A
Revision
21h
Sensor Status
Enabled
EVENT Output Control
Disabled
Temperature Accuracy
B-Grade
Temperature Resolution
0.2500°C (10-bit ADC)
Current Temperature
32.000°C
Negative Measurements
Supported
Interrupt Capability
Supported
10V of VHV on A0 pin
Supported




Frequency CAS RCD RP RAS RC FAW RRDS RRDL WR WTRS
1067 MHz 16 15 15 36 50 23 4 6 16 3
1067 MHz 15 15 15 36 50 23 4 6 16 3
933 MHz 14 13 13 31 44 20 4 5 14 3
933 MHz 13 13 13 31 44 20 4 5 14 3
800 MHz 12 11 11 27 38 17 3 5 12 2
800 MHz 11 11 11 27 38 17 3 5 12 2
667 MHz 10 10 10 22 32 14 3 4 10 2
Frequency CAS RCD RP RAS RC FAW RRDS RRDL
1600 MHz 14 14 14 34 48 39 6 8

 

[wingman99]

B-die great.I would try 3600 speed Ram 1.36-1.4v at 15-15-15-35.

 

[Brando]

do you mess with sub timings or leave on auto and do only main 4 timings with speed and voltage?

 

[wingman99]

I don't change sub timings they are left on AUTO. I set XMP, then 4 primary timings, speed and voltage. XMP does not seem to do anything on my Gigabyte with setting the memory manually, I just leave it enabled. Test the memory with 6 instances of MemTest HCI dividing the free memory you see in task manger with each instance of MemTest HCI.

 

[Brando]

Any ideas? I may have more voltage room to grow but i don't know what i'm actually at right now. I have it set to 1.38v in bios bit it shows 1.392v in hw monitor. Without the equipment to physically test voltage, how can i be sure if the mobo is over volting or just reading high? if the reading of 1.392v is accurate then either i got a weak set or i'm doing something else wrong. the vccio and vccsa look to be set sky high by xmp.

edit: sorry can't upload pic for some reason. settings are 3600mhz 15-15-15-35 at 1.38v (bios) showing 1.392v in hwmonitor. bumping voltage got me a bit further in the hci memtest but eventually got an error on 1 instance out of 6 at around 25-30% or so. this seems sub par compared to other reports of this ram but i could be messing up.

edit2: tried 3400mhz 14-14-14-34 @ 1.38v (1.392v in hwmonitor) and got an error around 10% on one instance of hci mem. i'm not good at this

 

[Brando]

ok im getting somewhere! 4000mhz 18-18-18-36 at 1.4v. just needed a hair more voltage from 1.392v to 1.4v. i can't seem to upload pics for some reason but it happened i swear!

 

[wingman99]

Cool your doing Great at this. If you don't mind I would like to know a few things. Did the ASRock work fine with using XMP and the new 3200 speed memory? When you tried 3600 15-15-15-35 did that work? When you find the final memory overclock what are your settings?

Some of the ASRocks have VRM stability problems with high voltage Vccsa Vccio. So if you have trouble try lowering Vccio to 1.1V vccsa to 1.15.

 

[Brando]

Cool your doing Great at this. If you don't mind I would like to know a few things. Did the ASRock work fine with using XMP and the new 3200 speed memory? When you tried 3600 15-15-15-35 did that work? When you find the final memory overclock what are your settings?

Some of the ASRocks have VRM stability problems with high voltage Vccsa Vccio. So if you have trouble try lowering Vccio to 1.1V vccsa to 1.15.

the ram worked fine out of the box with xmp on and nothing else at 3200mhz 14cas (stock) . i did get errors at 3600 cas15 1.392v but if i go back and try with 1.4v i'm pretty sure it will work since 1.4v seems to be enough for ddr4000 cas 17. it did really crank my vccsa and vccio to 1.27v and 1.35v at that speed but i lowered it a lot and will lower more on alarics advice. right now i'm rolling with 4000 cas17 since aida showed the best numbers that way and it seems to be stable. can't seem to post pics today. but heres an external link.

[url=https://postimg.cc/image/n666x4enl/] [/URL]

 

[Brando]

Can also check how high can you set memory clock and compare what is better. ASRock Z370 Extreme4 supposed to support 4333+. I left my memory @4500 after last tests and it runs like that for 3 weeks now

Btw. all new memory series are now on Micron and Hynix. I wouldn't be surprised if good Samsung IC disappear from the market or will be only in not many available 4000+ kits. In EU stores I can see only couple of 4000+ kits available. Almost only TridentZ and some overpriced Corsair.

4200 @ 18-18-18-36 1.4v. trying another bump in a minute.

how to screenshot on windows

 

[Brando]

and here is the highest i could boot without getting into the "dangerous" voltages. 4300mhz 19-19-19-41 @ 1.408v. not tested for stability whatsoever. should be able to go higher if i could grow a pair but im not exactly rich so my stuff needs to work for awhile.

 

[Woomack]

It will probably run like 4266 kits so 19-19-19 1.40V and IO/SA at auto but after tests try to set it lower. When you make it run stable then can play some more.

Some of the ASRocks have VRM stability problems with high voltage Vccsa Vccio. So if you have trouble try lowering Vccio to 1.1V vccsa to 1.15.

I had no problems up to 1.4V+ on all my asrocks but in last year I was testing almost only higher series (most X299/X399). At 1.45V+ many motherboards have limit or some OCP/OVP protection. On MSI from Z100/200 series, boards were shutting down on some BIOS releases because of too low protection for voltages.

You would think that the new Micron and Hynix would do better than Samsing ICs, why are they not doing better for overclocking? Why are the good Samsung ICs disappearing?

Matter of design, manufacturing process etc. Even Samsung didn't know how high their IC will run and I doubt they care much. They are not testing higher frequency when memory is designed for something lower. If it's specified for 2133 then they test 2133 ...
New Micron and Hynix are designed for higher frequency. All what you see now in Ballistix 2666+ or HyperX Fury or some other series is based on Micron E-die which runs at 2666. The highest IC in mass production is 3200 but there are downsides of that too. Lower frequency memory has starting timings also lower so higher frequency memory simply won't run at so tight timings because it's not designed for that.
There are timing tables in SPD and you can find them (usually, not always) on the IC manufacturer websites. Most Samsung B kits won't run at CL higher than 21. Most other IC designed for higher frequency will run at higher CL.
The same is in every generation so higher capacity - higher frequency - more relaxed timings. Some IC will still OC well, some not.
New Micron for me is quite good but because it runs at relaxed main timings then it's automaticly big meh on all popular hardware forums. Sub-timings run as tight as on Samsung, CL for 3600+ is 20-22 so it doesn't look so good... but performance isn't bad and can find it in cheap memory series.

Every IC has some period of manufacture. When on the market are appearing higher frequency or simply cheaper IC then older series are being discontinued. I think that the only discontinued IC which back to the market with the same specs and process was Winbond BH5 ... later changed to UTT.
Now we are in the point when some good batches of B-die were already binned and new batches are not as good. Maybe different factory or something else, hard to say. It's the same as some CPUs are better for OC, some are worse. Or Samsung is labeling a bit different chips with the same numbers, no idea.

 

[wingman99]

and here is the highest i could boot without getting into the "dangerous" voltages. 4300mhz 19-19-19-41 @ 1.408v. not tested for stability whatsoever. should be able to go higher if i could grow a pair but im not exactly rich so my stuff needs to work for awhile.

Looks like your loving the memory overclocking much better than the previous ASUS Rig. Glad you like the ASRock processor and memory overclocking. Memory overclocking is addictive, I started buying different Z370 manufactures motherboards to improve overclocking speed and stability.

Some of the ASRocks have VRM stability problems with high voltage Vccsa Vccio. So if you have trouble try lowering Vccio to 1.1V vccsa to 1.15.

I had no problems up to 1.4V+ on all my asrocks but in last year I was testing almost only higher series (most X299/X399). At 1.45V+ many motherboards have limit or some OCP/OVP protection. On MSI from Z100/200 series, boards were shutting down on some BIOS releases because of too low protection for voltages.

This is newegg with a few folks with Vccio and Vccsa VRM quality issue. link:https://www.newegg.com/Product/Prod...ignorebbr=1&IsFeedbackTab=true#scrollFullInfo
This is the ASRock forum I belong to with a few folks that have Vccio and Vccsa VRM quality issue. LINK: http://forum.asrock.com/forum_posts.asp?TID=6860&title=z370-extreme4-xmp-bug

Hello and sorry me again,

i found out the the xmp profile on the 1.3 bios for extreme 4 board has set
vccio and vccsa way to high. was set to 1.2 and 1.25 by default xmp.

For me this was the reason why the system could not boot into windows and was unstable.

I have now set xmp profile for my ram (3200) and then manually lowered

vccio to 1.1V and vccsa to 1.15. Not the system is rock stable again ram running finally with xmp speeds corretly.

Maybe this is helping one or another of you guys.

Best regards

flode

 

[Brando]

yeah idk what happened but i had 2 asus rigs in a row with memory instability and since i didn't really know ram very well it was a nightmare. maybe my fault somehow but all i know is i went asrock this time and everything jut works. thanks again for the recommendations and assistance. the old girl is so fast now i might have to give here a name. i did have a good time overclocking this ram kit as well. i didn't do crap yesterday except tinker and before i knew it, it was late. not sure i need to mess with it any more since i could really just leave it at xmp 3200mhz and not notice a difference but there's something about hitting that 4000mhz mark. feels good mayn. i bet that other 3600mhz cas 15 kit could do 4000mhz at cas16 though...

 

[wingman99]

It is not your fault in the past from what a remember all the things you tried. If i remember correctly you even tried getting new memory in the past? So many times in the forums we try to help folks to POST and run XMP stable or manually setting rated speed and nothing helps just like yours did in the past, I see it all the time. Mostly what I find in the forums XMP either just works and if does not tweaking settings manually when running the rated memory speed does not help.


It is sweet that the ASRock Z370s does 4000 speed with consistently in all the prices ranges compared ASUS with all my research I have done this year. From my research Asrock has the best Cores VRM phase layout and components in your price range and the best new t-topology memory tracer layout and shielding in all price ranges.

It does feel good to have the memory overclocked from 3200 to 3600+ even though the only thing I see a benefit in is synthetic benchmarks of raw memory transfer.

 

[Brando]

I think it's time to have a sticky in the memory forum called

"HAVING TROUBLE WITH XMP STABILITY? CLICK HERE!"