Search results

The difference between the MB you have tested that work and the one that doesnt could be nothing more that the settings pendragon suggested changing. The BIOS may be setting the V higher on one board than another or it may have more Vdrop than the other. It not really overclocking it is just...

In adition to the above make sure you have the latest BIOS.

Intel gets ~80GB/s multi threaded 40GB/s single threaded with that kit of RAM vs AMD who gets ~65GB/s multi threaded 60GB/s single threaded. So the website is just saying that anything that is much below 80GB/s multi threaded is under performing and mostly ignoring the latency and single...

👍 Looks like 5800X3D large cache adds a fair bit of memory latency so getting under 57ns is fairly good I would say and may be comparable to 5800X getting around 51ns. https://hothardware.com/reviews/amd-ryzen-7-5800x3d-review-and-benchmarks?page=2

Probably more to do with MB makers like GB tweaking the sub timings beyond what the kit is rated for. Gives a nice performance bump when it works but some sticks need a little extra juice to stabilize.

It's probably not completely stable with that small bump but it can make a difference.

That is a nice improvement for a 3d chip. I just noticed that you are only running 3 pass of TM5 you need to edit the .cfg file to increase that to around 24 to be a solid stability test. Also not sure if you missed this late edit which may help if you're current settings prove unstable...

If temps were high I can guarantee you would have errors at 240 tRFC and tRC 40. Oh also try tRDWR 7 or 9 if that fails.

Would be interesting to see gdm disabled latency test even if it needs some extra juice just for the test. Extra v is only a problem from a heat stability point of view as the RAM can take it. You may be able to lower ProcOdt a little. With GDM disabled only you could try tWR 10 with tRTP 5...

A little more detail on those choices. tRCDWR 9 (I think this gives you a tRCD value of 12 for calculating tRAS once averaged with tRCDRD 15) tRAS 27 (I remember it being suggested to add +2 to the combined value of tRCD and tRP)

Lower your VDDG IOD to ~0.07V bellow VSOC and the VDDG CCD to ~0.05V bellow IOD and see if you can drop the CLDO VDDP back to 900 without loosing stability. VDDG draws it's power from the VSOC so it cant be higher and running the same will result in Vdrop hence why it is best to keep it lower...

3800C14 is a tall ask and most sticks will need over 1.5v unless you got very lucky. I have 8x8GB sticks of b die 4x3200c14 and 4x4400c19 and only two of them can do 3800c14 with under 1.5V. The settings you are following have VSCO lower than VDDG which is not ideal so it may not be the best...

From memory 2 is best for two single rank sticks and SCL 4 is usually best for 4 single or two dual rank. But it has been a while since I was overclocking RAM and there is just to much to remember, so try both and see if there is a performance or stability difference unlessriev90 can confirm...

Since you have dual rank RAM you want to run RttNom 6, RttWr 3 RttPark 3 and try lower ProcOdt, these settings should improve stability. You could also try VGDG IOD ~1.050-1.060V but if memory serves this is more for pushing past 3800. Try drop the SCL to 4. You can find a excellent guide on...

Everything else looks to be much the same.

Kingston is a easy choice then as it has a slight speeed advantage with no down side.

The kingston kit is the fastest but depending on price it's probably not the best value.

3200CL22 13.75ns 2666CL18 13.5ns 2400CL15 12.5ns So there is a very small latency loss for a fair increase in bandwidth. If you dont want to lose any latency you could find a kit of Kingston FURY Impact 3200CL20 12.5ns Same latency as 2400CL15 due to the frequency increase offsetting the timing...

It will be negligible due to the timings going up at a similar rate to the frequency making the end latency much the same but with a little more bandwidth which would benefit the IGP but not make much difference to the CPU. So not really anything to do with the Intel arch in my mind. Fair...

With 3770K, 6700K, 7700K, I found that overclocking RAM with ~1.5Vdimm made a larger difference to performance in most games than overclocking the CPU and putting both together made for a very noticeable improvement. But I guess you could say that each on there own was negligible at ~7% from a...

This is a myth Intel sees gains from latency\bandwidth just like AMD does. In the past RAM speed has been enough to make i3 NUC faster than i5 NUC in tasks that are sensitive to it. Nice to see that Intel has added XMP support to NUC again after removing it for a few generations.

How different is your kit from your mates? maybe he would swap with you. Otherwise maybe it is time to replace your kit with a 3200c16 2x16GB kit and sell your current sticks.

My guess is the BIOS is setting the RAM sub timings more aggressively for the stronger memory controller on the new CPU and it doesn't like it. Any chance you can get one stick on it's own to post? if so you may be able to try setting 3000 1.4vdimm manually and see what happens when you plug the...

It would be interesting to see a AIDA64 latency benchmark at 1:1 / 1:2 / 1:4 with DDR5 at 4200.

As far as I can see DDR5 can't be set to gear 1, do you have any links to proof that it can? The memory controller and MB traces are simply not good enough for 4 sticks of RAM. If there was a 10 layer T topology MB with two RAM sticks on the front of the PCB and two on the back it would do a bit...

If you want to get dual rank 16GB sticks Kingston provides specs for there kits which tell you how many ranks they have. Or there is the expensive option b die which is not really worth it now that DDR5 is out.

I am running 4x8GB b-die and have a great OC with a 5800X but if I had 2x16GB high bin b die and a MB with two RAM slots it would probably do a little better due to the shorter traces to the RAM. I went with 4x8GB due to price. 4x8GB b die and 2x16GB b die are both dual rank while most new...

Testing one stick at a time is how you confirm that the RAM is at fault as it is very unlikely for both sticks to die at the exact same time unless they had some help.

Looks like good value at $149us The CL actually means very little to DDR5 performance, it is more about frequency and sub timings which are not advertised and often left to the MB to decide. This 6800 kit for $210 may be worth the extra to someone looking at spending a bit on a expensive CPU...

Probably because your JEDEC profile for the kit is 4800 check with CPU-z on the SPD tab, it will show the profiles with JEDEC being the default. Corsair has fallen behind as far as RAM goes and there is a bunch of brands I would take over them. G.Skill having MB QVL on there site is great and I...

Well you are and you are not. The actual memory chips are overclocked but they are tested and warranted to that speed by the company that assembles the stick (This includes 5600 sticks) A high speed RAM kit shouldn't need any tweaking with a half decent Z790 MB and a 13 series CPU all you need...

13 series CPU will have a better chance than 12 series so helpfully you also have that. How far you can OC depends on the exact model of the kit and what memory chips it is using. But even if you don't have a kit that can hit high MHz you can still see huge gains by tweaking the sub timings...

Look at your RAM usage in task manager performance (CTRL SHIFT ESC) If you are using 16GB+ and are seeing a bit of disk activity while you work then more RAM could be a huge improvement. But if not then it will make no difference.

With 32GB next to 8GB you would only have dual channel for the first 16GB of data while the remaining 24GB will be accessed at single channel speeds. That is if the MB supports running dual channel with different sized sticks. If not then it will just be single channel all the way. With 16, 16...

Memtest will let you know if one of the sticks is faulty and if so try one stick at a time till you find the culprit. But it probably wont show errors if the memory controller is unstable due to the different sticks of RAM, for this you want to use OCCT (CPU, large data set)

If you are looking to OC and are chasing high frequencies on Intel then newer sticks with Hynix A die are what you want.

Thaiphoon burner or a look at the PCB will tell you what you have. Patriot Viper 4400c19 is a cheap A2 but not the best bin of b die so it is hit and miss with how tight you can get the timings. I had a lot of trouble with tRCD on a few of my b die sticks getting negative scaling at high v with...

Depends on the PCB. A2 PCB can do ~4400+ while others can struggle to go much over ~4200 but can handle tighter timings better.

Seems I was wrong. 6400-7200 is seeing up to 11% increase in CPU performance here. I cant help but wonder if your 6400 kit could OC to 7200+ though.

Higher RAM speed helps CPU performance. Resolution has virtual no impact on CPU performance just GPU except for some older game engines. So if you are playing a very CPU heavy game that makes GPU usage drop when you get low FPS dips then faster RAM may help. But just buying a faster kit wont do...