• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** AMD "Zen 4" thread (inc AM5/APU discussion) ***

Remeber all it will do is save you time and effort setting memory frequency, timings and voltages yourself. So in theory should be able to buy a non EXPO kit, enter the BIOS and set all these yourself. As noted below, XMP profiles will work in the same way and the kit will likely have some of those which you can apply instead. The difference between them and EXPO is that EXPO has been certified stable for Ryzen 7000 series, where as XMP have not.
Ok mate cool, thank you. So in theory I could potentially use the XMP profile and it would be ok? I've never had to set timings and stuff before, is it hard to do? Difficult to get working? All the reviews I've seen when they put ram in the machine takes a while to load up because it's trying to get the best timings, is that EXPO?
 
Cheers Gibbo.

@cHk4 That's probably a no then. The AMD EXPO profile are stored on the DIMMS, so unless they have the AMD EXPO branding then it won't have them. However, it may have alternative profiles which can be used. That said, this is just speculation on my part and I don't have any non AMD EXPO DIMMs to test with unfortunately.


Using EXPO/XMP memory is considered overclocking. Overclocking voids warranty, this is not new and this would apply anywhere. No one filing a warranty claim is going to admit they overclocked, well I guess there's always one person that might. Therefore, warranty claims should get processed just fine so long as the claim is legitimate. This has been the same on previous gens too, it's not new.

Reviewers were asked to test using two identical kits, running at the same frequency for fair comparisons Vs Intel processors. Didn't want to have the AMD processors running a faster kit and the Intel stuck at 4800. This is why you see places like HUB backing up the results AMD showed, as although they used even faster memory for Alder Lake (6400) the differences were not huge.

Meanwhile, at the Intel event showing off Raptor lake, Ryzen processors were stuck using 3200Mhz, when its widely known that Zen3 sees big improvements running at 3600-3800Mhz CL14-16. Alder Lake was also running 4800Mhz memory vs Raptor Lake at 5600. What do you think about that?

RPL baseline has been moved from 4400 to 5600 so all three system have been baselined in their vendor approved configurations.

At the launch of ADL, Intel shared all their official data with with 4400 kits even though they would have greatly benefited from the performance of 6000+ kits that were already available.

In the same way they used to use Jedec baseline at the launch of the 9900k when *much* better options were available for generations and as we all know, scaled amazingly well. https://newsroom.intel.com/news-rel...or-new-9th-gen-intel-core-i9-9900k/#gs.dfcdep

Hopefully actual data and information can avoid a new conspiracy.....
 
Last edited:

This board supports an add in card for Thunderbolt 4, wonder if that would work out the cheapest?

Its still a little confusing to me, all i know is Zen 4 has 28 PCIe 5 lanes, 4 of them are for the chipset so 24 'Usable' 16 for one GPU or 2X 8 if you're running SLI / CF, 4 lanes for NVMe storage X2.

So one thing we can say from that is it doesn't steal PCIe lanes from the GPU if you run NVMe drives, Alderlake only has 16 usable PCIe 5 lanes, whether or not it will use the fastest lanes in whatever PCIe slot you plug something in to i don't know.
 
Last edited:
Ok mate cool, thank you. So in theory I could potentially use the XMP profile and it would be ok? I've never had to set timings and stuff before, is it hard to do? Difficult to get working? All the reviews I've seen when they put ram in the machine takes a while to load up because it's trying to get the best timings, is that EXPO?
Yes, however stability would not have been verified, so you would need to check that yourself.

EXPO is definitely not meaningless, so I would not listen to that but it is not essential.

More information here: https://www.amd.com/en/technologies/expo#
 
Last edited:
These are all the officially supported frequencies though?

I think it really depends on what is being tested.

If you are doing a science app focused test for a work environment then sticking to spec makes sense.

If you are doing a general test or gaming tests then using the best reasonably affordable ram that works is probably the way to go.
 
I get the point, but if the GPU is not strong enough then it shortens the gap between the CPUs. As the whole purpose is to test CPU performance, you want the fastest GPU for at least the initial performance review.

This goes someway to at least explain the difference in results of TPU vs everyone else, and maybe there are other factors too tbf.
Indeed but I did find it useful to see that honestly if I am looking to see if I need to upgrade based on current GPU (that I am sticking with) then it would be wasted and I would be better off saving for a new GPU next gen than CPU right now and then look at CPU the following generation as example. If my GPU is a bottleneck regardless then the GPU needs to be my first upgrade. That will be better to understand come having the 4080 12GB 4060 and 4090 to show differences.
 
Hi there

If there is any particular DDR5 kit customers are after made by Corsair, Kingston or TeamGroup, let me know and I shall attempt to source it. :)
Legend! Do you know which EXPO kits are arriving first and when? I'm literally on the edge of ordering everything but don't know about the RAM lol :D
 
Most the Corsair EXPO stuff should be with us in the next 1-3 days and Kingston lines 3-5 days.
Thanks mate. Maybe one final stupid question, I've been told numerous times to check the QVL for the board I'm getting (ROG Crosshair x670E Hero) but none of the Corsair ram is listed there, only the old stuff. I know you don't work for ASUS but I'm guessing that is due to the fact that the RAM hasn't been released yet?
 
Thanks mate. Maybe one final stupid question, I've been told numerous times to check the QVL for the board I'm getting (ROG Crosshair x670E Hero) but none of the Corsair ram is listed there, only the old stuff. I know you don't work for ASUS but I'm guessing that is due to the fact that the RAM hasn't been released yet?

Pretty much be that yes, I think in these early days people will need to be manually setting up memory kits no matter what they buy, just stick to 5200-6000MHz and all should be fine. :)
 
Welp I've done it -

YOUR ORDER HAS BEEN PLACED SUCCESSFULLY!​


Scared to death of going AMD but there we have it, I'm going to the red side... In before I get my new ram and I start spamming the RAM section because I'm clueless :D

Now need to cancel my 13900k... or do I?
 
Thanks mate. Maybe one final stupid question, I've been told numerous times to check the QVL for the board I'm getting (ROG Crosshair x670E Hero) but none of the Corsair ram is listed there, only the old stuff. I know you don't work for ASUS but I'm guessing that is due to the fact that the RAM hasn't been released yet?

QVL serves two main purposes from a practical standpoint. One is to get an idea of what the board can do in the different memory configurations. For example, with all 4 slots populated you'll have a lower frequency ceiling than with just two. You can use QVL to determine that delta. Secondly, when new IC's come out, you want to ensure they're on the QVL list which means the mobo manufacturer has updated their training algo's to work with that IC. This can often require a bios update as they have to get the new training to you somehow.

So if they've tested Hynix kits at 6400mhz already and you get a Hynix Expo kit that's 6200 but not on QVL, that doesn't mean it won't work. There's some complexity around different IC's from the same manufacturer but that's sidetracking a bit.
 
Last edited:
QVL serves two main purposes from a practical standpoint. One is to get an idea of what the board can do in the different memory configurations. For example, with all 4 slots populated you'll have a lower frequency ceiling than with just two. You can use QVL to determine that delta. Secondly, when new IC's come out, you want to ensure they're on the QVL list which means the mobo manufacturer has updated their training algo's to work with that IC. This can often require a bios update as they have to get the new training to you somehow.

So if they've test say Hynix kits at 6400mhz already and you get a Hynix Expo kit that's 6200 but not on QVL, that doesn't mean it won't work. There's some complexity around different IC's from the same manufacturer but that's sidetracking a bit.
That's great mate, thank you. I didn't realise that about the 4 slots, that makes sense though. Appreciate the reply :D
 
Pretty much be that yes, I think in these early days people will need to be manually setting up memory kits no matter what they buy, just stick to 5200-6000MHz and all should be fine. :)
Ordered at 12:30pm Gibbo, I hope it'll be dispatched today in time for the weekend? :)

Welp I've done it -

YOUR ORDER HAS BEEN PLACED SUCCESSFULLY!​


Scared to death of going AMD but there we have it, I'm going to the red side... In before I get my new ram and I start spamming the RAM section because I'm clueless :D

Now need to cancel my 13900k... or do I?
You'll be fine and plenty of folk round here that can help if you need it. :)
 
Last edited:
Do you have a link to that review please for context?

It looks like the Intel processors lose performance quicker than the Ryzen processors.
They dont lose performance, they drop to 125w. You know, the thing ive been saying since yesterday.

I think puget uses a u12a, the same cooler i have. It doesn't thermal throttle with a 12900k (although it might with the ks, dunno), so what you are seeing must be tau expiring.
 
Back
Top Bottom