12900K + 12600K DDR4 vs DDR5 – Surprising Results!
Finally, I have my hands on a Z690 motherboard that supports DDR4! It’s taken a little while, but I’m really glad to have this in because the results aren’t quite what I expected! Both chips seem to perform slightly differently with the different types of memory, so let me walk you through my results.
First off as a brief refresher, these are Intel’s newest generation of CPUs, codenamed Alder Lake. Specifically I’m using an i5 12600K and an i9 12900K, and among the variety of significant changes including two different types of CPU cores, is support for both the ‘current’ generation memory called DDR4 (double data rate gen 4) AND the incoming ‘new’ generation called DDR5 (double data rate gen 5). DDR4 and DDR5 have some significant differences I explained in a video recently, one I’ll link in the cards above for you check out after this one, but suffice to say they aren’t cross-compatible – ie you can’t run DDR4 and DDR5 at the same time, they don’t even fit in the same slots. That means you need to purchase a motherboard that explicitly supports one or the other, with most motherboard vendors (upon Intel’s recommendation) picking DDR5 to be the default, and offering ‘special’ DDR4 versions of a small selection of their lineup that seem to be commonly denoted as either “DDR4” or “D4” models.
And that’s what I’ve got here, an Asus STRIX Z690-A Gaming WiFi D4 board. This still isn’t exactly cheap, currently listed for around £320 on OCUK, but it does seem like the DDR4 versions of boards are generally a touch cheaper than their DDR5 counterparts. Add to that the cost and outright inability to purchase DDR5 and you’re left with not so much a decision as a requirement. But how does DDR4 actually perform with these new chips? It must be terrible if Intel say you probably shouldn’t use it, right?
Well, looking at the Puget Bench results from After Effects you’d sure think so! The 12600K using DDR4 scores almost 14% slower than the same chip using DDR5! The i9 is similar, although less severe with a 5% performance loss when using DDR4, but a loss is a loss. Premiere is the same story, almost 10% less performance from the i5 and 6.5% less from the i9. Photoshop is a little interesting though, with the i5 reporting 11% less performance, but the i9 reporting almost 10% MORE.. A bit odd, I know!
It remains… confusing in Blender as in the BMW scene the i5 drops 1.3% when using DDR4 instead of DDR5, but the i9 matches it’s original score fully. Neither changes the order of the chart here though, as even the i5 is only 2 seconds slower to render the frame, with the next fastest chip being the 10850K running a further 6 seconds back on that. In Gooseberry though, the DDR4 equipped i9 actually runs marginally faster, while the i5 actually drops a place, albeit only by 1% slower than the DDR5 result.
Cinebench R23 is also confusing as in the single threaded results, the i5’s DDR4 run was around 2% faster than it’s DDR5 run, but on the i9 it was the other way around, with it dropping around 1%. But, in multi-threaded, both the i5 and i9 drop between 1 and 2% on their respective results. The good news here though is that beyond the Adobe suite apps, the performance differences don’t seem to be too massive, and even in the Adobe apps it doesn’t look like it’s such a significant difference that it would drastically affect your usage experience.
Ok, so on the whole it doesn’t look so bad, but what about the all-important gaming results? Well that’s rather mixed too. Microsoft Flight, especially on the i9, offered more performance using DDR4 than 5, with a significant improvement of over 10% on the 12900K but a more measured 2% on the 12600K instead.
Fortnite is pretty much within margin of error on both chips, within one or two FPS which at 260-270 FPS average is perfectly fine. Watchdogs on the other hand shows a significant hit on both chips, netting 12% less performance on the i5, and not far off 20% less on the i9. Even the 1% low figures drop by over 10% on both chips. While it’s of course still perfectly playable it’s far from ideal, as it erodes the lead these new chips have over AMD’s Ryzen CPUs and even previous generations on Intel CPUs.
Cyberpunk also suffers, with around 8% less performance from the i9, and almost 15% less from the i5 which is enough to take it from the top of the pack to more solidly mid-field. Interestingly though, the 1% low figures actually remain relatively high compared to the average FPS meaning despite basically matching the 5900X on average, you’ll likely have a slightly better actual experience playing on the i5 with DDR4.
Lastly in CSGO, both chips drop places compared to their peers with the i5 taking a much larger dive. The 12900K slips behind the remaining 5th gen Ryzen CPUs I’ve tested, if only just, but the 12600K drops almost 100 FPS average taking it to the back of the pack, save for the last gen 11600K. Of course I’m talking about 500 to 400 FPS here so you will not perceive that difference at all, but it’s pretty interesting for me to see that much of a swing.
Looking at just those two, setting the DDR5 run to 100%, you can see on average the i9 drops around 4% worth of performance across these games. By comparison, the i5 drops more like 7% across both average and 1% low figures, or more like 9% if you just look at the average FPS results. While that isn’t a complete deal breaker, it does change the price-to-performance ratio, especially considering the added cost of the motherboard. Even a £120 B550 can support any of the Ryzen CPUs relatively well, and a more reasonable £150-180 board will be more than enough, compared to the £260 for what should be a relatively budget class board or the £320 for this one.
Taking the current pricing of all of these chips, CPU alone, using DDR4 actually drops i5’s the FPS per pound to almost a match for the Ryzen 5800X, rather than rivaling the much cheaper 5600X with DDR5. The new i9 sits comfortably at the bottom of the chart here, but even so using DDR4 sinks it even lower.
If you factor in the motherboard and RAM cost, which I went for £130 for any DDR4 board and £300 for DDR5 as that’s the last price I’ve actually seen it listed for, and went with £180 for the B550 Ryzen boards, £210 for Z590 and £320 for either Z690 board, you get a rather interesting comparison. The only Intel chip that comes close is the last gen 11600K, which is only just ahead of the 5900X and noticeably behind both the 5800X and 5600X. The 12600K with DDR4 does come out ahead compared to using DDR5, but not by all that much. Going with a cheaper board would definitely help there, but the same could be said for any of these options. From a sheer price perspective, it’s hard to argue these new chips are exceptional value.
