• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FEATURED Patriot Signature 2x32GB DDR4-3200 CL22 SODIMM - PSD432G32002S (inc.tweaking)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.


Benching Team Leader
Jan 2, 2005
Today I will present something a bit different so 64GB DDR4-3200 SODIMM memory from Patriot. Since the gaming series wasn't released yet (will be soon) then I started to play with the Signature Line series. In short, it's standard series memory designed for maximum compatibility and stability rather than maximum speed. It doesn't change the fact that for laptops it's still one of the fastest options but is simply using JEDEC specification for DDR4-3200 so gamers would say that timings are really relaxed.

Some photos of the Patriot Signature SODIMM memory.





The tested memory kit contains two 32GB memory modules packed separately. As I mentioned, the memory has only an SPD profile with a higher JEDEC profile so DDR4-3200 CL22-22-22-52 and 1.20V. Since the SPD profile doesn't have the option to use any higher voltage then any overclocking or timing adjustment is a bit tricky.
The memory is based on Hynix M-die so about the same as we can find in many popular desktop 32GB DDR4 modules. Memory chips are rebranded but Thaiphoon Burner is showing us the IC details.


It also works fine at default settings


The memory has been tested on the ASUS TUF A15 laptop with Ryzen 7 4800H processor and GTX1660Ti graphics card. This laptop and probably also other laptops with Ryzen 4000 processors are limited to the DDR4-3200 memory clock and there is no adjustment in BIOS. To make memory faster I had to edit the SPD profile and after a couple of hours, I managed to set a fully stable DDR4-3200 CL18-19-19 profile which already looks good considering that in use are 32GB modules and even the best Samsung B-die won't really run below CL18 at 1.20V on this platform (I actually tested that with another kit based on Samsung B).
Last edited:
Since we are limited to DDR4-3200 and everything at the lower clock will perform worse regardless of timings then the only thing we can do is to play with timings. Laptops won't have memory timings adjustment so with help comes Thaiphoon Burner software which can be downloaded from softnology.biz website. Editing SPD and XMP profiles aren't available in the free version so any modifications will cost us but thankfully not so much.

I'm usually showing many various settings but because of the mentioned limitations here are only two results, at default settings and with adjusted timings. It's more to show what this memory can do than directly compare the performance.
In games, you can count on anything between 0 and 10FPS increase so about 0-5% performance gain. For example, Tomb Raider or Far Cry 5 at 1080p in built-in benchmarks are showing up to 7FPS more between DDR4-2666 CL18 and DDR4-3200 CL18 and up to 3FPS difference between DDR4-3200 CL22 and DDR4-3200 CL18. Depends on settings and display resolution it can be more or less.

SPD DDR4-3200 CL22-22-22-52 1.20V


SPD DDR4-3200 CL18-19-19-39 1.20V



The SPD profile after timing modification. Fully stable settings as on the above screenshot. The memory works like that for two weeks without issues. I also know that Patriot will soon release the gaming series SODIMM at CL18 so it should perform about the same with 32GB modules. Lower capacity modules are usually faster because it's not possible to use all timings in the SPD profile and a long list of sub-timings is adjusted by the motherboard/BIOS.

Results are in short about 2GB/s higher memory bandwidth and 9ns lower latency. As far as we won't see a significant difference because of the memory bandwidth then the latency is what makes some applications work slightly faster, what we can see in games.


I hope it will help someone in the future. This is only a solution to make a memory in the laptop faster. You have to answer yourself if it's worth the time and risk.
Thaiphoon Burner is a great tool if used well. There is always some risk to damage something if you don't know how to do it.
Last edited:
I always enjoy reading your posts on memory. You seem to leave no stone unturned
Thanks, I'm glad that someone is reading my tests/reviews :) ... but really, there are a lot of readers but barely any comments. If anyone wants any specific tests then can always ask.

Since I purchased a new laptop then I asked some vendors to send SODIMM review samples. Not many actually have DDR4-3200 SODIMM in their offer. Most have only up to DDR4-3000 or DDR4-3200 but with XMP profiles that are not working on most laptops. I have Corsair DDR4-3800 SODIMM that I was using on ASRock X299E-ITX/ac but even though seems like much faster memory than on my laptop is also limited to DDR4-3200 CL18 and SPD is only DDR4-2666 so I had to modify the profile to check how high it will go. If I have to pick then 2x32GB is a better option than 2x8GB, even though I won't really use 64GB but when I edit some graphic files then 16GB is already not enough.

Btw. Patriot just released higher frequency Blackout kits, up to DDR4-4400. One of these kits should be on the way to me so I will show some results. I'm just not sure when as I wanted to post 2 more threads about memory kits that I have for 1-2 months, and so far I had no time to gather all up and do additional tests. With desktop Ryzen 4000 premiere, I will for sure show something more.
I have a kit of the Patriot Blackout 4000 mhz with samsung b die and am happy with them for the desktop
I guess that new kits have some other IC. Patriot generally uses Hynix IC and most brands make higher frequency kits on Hynix DJR or Micron B/E right now. I think it will be Hynix but I would be happy with the one that can reach DDR4-5000+. My last 2 or 3 kits couldn't pass DDR4-4800. With Ryzen 4000 premiere should also come better memory controllers and higher frequency for 1:1 IF ratio.
Want to test lower timings

Laptop: Asus ROG Strix G17 G713QR-HG022T (Ryzen 5900HX, RTX 3070 130W) that came with 2x8GB 3200MHz CL22 1.2v Samsung memory.
That kit has been replaced with G.Skill RipJaws 3200MHz CL22 1.2v 2x16GB (F4-3200C22D-32GRS) and now memory is way faster, it has memory modules on both sides, the 2x8GB sticks had only on one side. In Warzone I get roughly 15% more FPS with dual sided modules.

Mobo doesn't allow any ocing, there are no options at all, so I found this thread by @woomak about changing timings, I'm very curious about it but wanna know more before I go and do my first attempts.

The RipJaws have a series of SO-DIMM 3200MHz, CL16, CL18 and my CL22, could it be possible to get mine to CL19-20?

Why would my mobo accept CL20 with the tweaks that program does, when I see people failing to get CL18-CL20 working (mobo sets CL22 with their CL18 or CL20 memory) without the tweaks? Is the information somehow stored on the RAM or how does it work?

Edit: Btw, since two BIOSes back, SPD information is somehow unavailable, CPU-Z and AIDA64 Extreme show nothing, all blank.
Downloaded Thaiphoon Burner, it can't read SPD either.

Thankful for advice :)
Last edited:
Most these laptops don't have XMP option so you are limited to 1.20V. At 1.20V you can use only "standard" modules like 3200 CL22-22-22. The highest ones at 1.20V are sold at CL20 if I'm right. The "standard" profile is programmed in the SPD so regardless of anything else, as long as the motherboard supports memory clock then will use the SPD profile.
If you see DDR4-3200 CL16/18 and most CL20 too, then probably run like that at XMP settings and 1.35V. So again, it won't work on most laptops.
I was testing multiple 2x8GB, 2x16GB and 2x32GB SODIMM. At 1.20V I couldn't make then run below CL18. The lowest were Samsung B at 3200 18-18-18 or 18-17-17. However, as you said, dual-rank is a bit faster and then 2x16GB or 2x32GB can make not much worse main timings (but worse secondary). I could set 18-19-19 stable and the memory in this thread still works like that in my laptop. On the other hand, Team Group Zeus that I recently tested, with the same IC, runs at 3200 CL20-20-20 and nothing below that. Silicon Power 2x16GB could make 18-20-20 as I remember. I assume you can set anything between CL18-18-18 to 20-20-20, depends on used IC and luck.
SPD should be available in popular software. In CPU-Z check 2-4 tabs from the list, 1-3 are usually empty.
Thaiphoon Burner can check if SPD is write protected or not. Can do that with free version before spending money. On the other hand, I don't think that improving memory profile will help you much in performance.
Just wanted to show the comparison in memory speed using UserBenchMark app, original 2x8GB dual channel with modules on one side vs aftermarket 2x16GB dual channel with modules on both sides.
AIDA64 Extreme doesn't give separate single/multi core results. Normal AIDA64 bench: Read is 47K, Write 43K (very very few times I've seen 45K, and only once 47K), Copy 42K


The RipJaws 3200MHz (2x8, 2x16, 2x32GB) SO-DIMM series are CL16 (XMP), CL18 (XMP) and CL22 (no XMP I guess) - all stated 1.2v
If RAM has XMP, it defauls to your mobos settings (CL22-22-22-22) no matter what Thaiphoon Burner can change?
Meaning as a laptop user (AMD system, no XMP enabling in BIOS) has best chance with non XMP RAM?
I had the impression that having 3200MHz 1.2v CL16 could give me the lowest timings editing with Thaiphoon Burner - but that's a no go if XMP is on the RAM? Can it be "erased/changed" from the sticks?
Last edited:
Nowadays, nothing is single core. Windows is balancing the load and uses all cores. Check games or anything else during work and you will see that single cores are never 100% loaded for longer.
Another thing is that this benchmark says nothing to me. It's more like a marketing tool than a real comparison.

If RAM has XMP but the motherboard can't use XMP then the motherboard uses SPD profile. Since you can set memory voltage only manually or via XMP and laptops, in general, don't support manual adjustments or XMP then you have no big choice but to use RAM with programmed SPD at a higher frequency and tighter than standard timings.

SPD and XMP can exist together. It's just like I said, if XMP isn't supported, then the motherboard uses SPD. So no matter if there is XMP or not, it won't be used. Another thing is that XMP has to be enabled manually and it won't load automatically at first start.

I think that even G.Skill is not selling DDR4-3200 CL18 1.20V anymore. For me, it's a bit weird as DDR4-3600 SODIMM were available a couple of years ago but not anymore and new platforms are not even supporting more than DDR4-3200. XMP was available in more gaming series laptops but right now you can barely find it anywhere.
I play plenty of modern titles where one/two (perhaps a few) cores are pegged 100% and the rest show minimal use/idle. In fact, I'd venture to say more games are like that than actually utilize all cores and threads. COD is the only one I can think of that uses all the cores/threads it can (not even talking my 18c CPU, but 8/10c parts).

+1 on userbenchmark being BLEH.
these are AIDA64 bench comparisons, 2x16 vs 2x8GB on my laptop

Memory Read +11.2%
Memory Write +34.7% (33535.8 GB/s, 45178.8 GB/s)
Memory Copy +12.4%
Memory Latency 4.4% lower
CPU PhotoWorxx +36.8%

(5 runs to get an average to compare with)

I played Warzone with HWiNFO showing DRAM Bandwidth, my laptop showed 14Gbit/s READ, and maybe 5Gbit/s WRITE, how could 11.2% increase in memory read speed give 15% more fps i n Warzone?
Maybe all is in Memory Write, Multi core? *shrugs*

Seems I trust UserBenchMark Multi score more than AIDA64's general.

I found this own note too:
42% faster Multi Core Write in UserBenchMark
Last edited:
these are AIDA64 bench comparisons, 2x16 vs 2x8GB on my laptop

Memory Read +11.2%
Memory Write +34.7% (33535.8 GB/s, 45178.8 GB/s)
Memory Copy +12.4%
Memory Latency 4.4% lower
CPU PhotoWorxx +36.8%

(5 runs to get an average to compare with)

I played Warzone with HWiNFO showing DRAM Bandwidth, my laptop showed 14Gbit/s READ, and maybe 5Gbit/s WRITE, how could 11.2% increase in memory read speed give 15% more fps i n Warzone?
Maybe all is in Memory Write, Multi core? *shrugs*

Seems I trust UserBenchMark Multi score more than AIDA64's general.

I found this own note too:
42% faster Multi Core Write in UserBenchMark

AIDA64 memory and cache benchmark is comparing only pure memory/cache bandwidth and latency. To see how Windows uses memory, try winsat mem command from command prompt. It will give you results close to AIDA64 memory copy results. Also, in AIDA64 when you hover on the icon next to read/write/copy results then it will explain that memory copy is actually the result that translates into the daily performance more than anything else.
There are many variables that affect general performance. No one is saying that dual rank isn't faster as on new Ryzen CPUs it is faster and it was confirmed by many sources. In short, your results are correct so just enjoy having better performance on the 2x16GB kit. A bit higher performance and a quite low price of RAM made me use 2x32GB in my laptop, even though I probably never use more than ~20GB.