Earlier today, AMD announced the fourth generation Ryzen lineup, called Ryzen 5000 (they did it to themselves and it is easier than trying to explain the differences in 3 product lines all called Ryzen 4000). I wrote the broad strokes of that news already earlier today, so I want to zoom in very specifically on a point or two of contention I have with the news of this lineup.
Firstly and less substantial, the performance. The performance uplift claimed is an average of 26% over the last generation when comparing just the 12 core parts in each lineup to each other. This is great news, but it also feels like the best result cherrypicked from the pack. That’s a fairly standard practice (look at how Nvidia compares across generations, for one) but still worthy of calling out. It is difficult to say if that increase will remain in place across the stack, like 3800x vs 5800x, or the 6-core variants of both parts being compared. In some ways, I expect the 6-core comparison to be better for the fourth-generation parts, because the clockspeed increase is larger proportionally! The uplift against Intel is the interesting part to me, as that is the thing I most wanted to see as an AMD enthusiast, and it looks good, but also not amazing in all aspects. The massive increases to League of Legends and Counter-Strike: Global Offensive, games that are heavily CPU dependent, is outstanding, but the slide shows a much less impressive set of wins that are mostly single-digits or within the margin-of-error, and the one loss in Battlefield V. (As a sidenote, this is why I trust AMD slides for CPU more than Nvidia’s – including the negative is a show of faith when they could have just left it out).
That isn’t to say that this is bad or not a sufficient performance uplift – I expected less and got this, so as someone wanting to purchase a new system with Ryzen again, this is great news. However, a smidge of skepticism remains – I want to see more benchmarks and independent benchmarks done. While the test at 1080p does mean the CPU gets stressed more, it also means that for many enthusiasts who have moved on to 1440p or higher resolutions that the performance uplift is likely to matter less. It should, however, make AMD a viable choice for high resoution, high refresh rate gaming, like 1440p 144Hz or even the slow start of high refresh rate 4k panels hitting the market. I’d be very curious to see how CS:GO in particular scales up as resolution increases, since the numbers are often so comically high that they set apart CPUs due to visible deltas in performance.
That leads me to the pricing, however. To be clear, I don’t think a price increase is inherently bad, but one must be justified by suitable improvements in performance. I think AMD is on the cusp of being able to claim that, but the increases across the product stack are baffling. Offering the 6-core model at $300 when your current generation 8-core part is $330 at full retail price and rarely sells that high anymore is a highly confusing decision.The Ryzen 7 3700x can be had today for $295, and while that does mean a decrease of 19% in IPC, you gain 33% the cores. Sure, they have the architectural deficits that Zen 2 has in comparison to Zen 3, but those also factor into the IPC increase as-is! That to me is a huge weak spot in the lineup, because the 6-core parts being sub-$250 and having a sub-$200 non-X SKU helped break AMD into the market and is money left on the table.
When I put it like that, though, it kind of makes sense. If I’m AMD, and I’ve produced a large supply of 6 and 8-core CPUs in my current lineup that remain available in retail, does it make sense to push people away from those parts to my new ones? Perhaps not.
On the other hand, Nvidia manages this more gracefully in the GPU market – when Pascal cards at the mid-range were overproduced, they let them sell through by delaying the launch of Turing until a majority of the cards were sold. GTX 1060s were popular choices during the mining craze as they maintained a semi-sane price and ended up overproduced as a result, so Nvidia just pushed the whole Turing launch back a few months (allegedly) and then didn’t introduce the RTX 2060 until January 2019.
The thing about increasing the 6-core’s pricing so much is that it doesn’t make a lot of sense to buy at launch. The staggered launch rumor made sense to me because it would see AMD push the 6 and 16 core launches late in the year, allowing the existing stock with some markdowns to clear shelves before introducing them. However, that isn’t what we’re getting. On top of all of that, only the 6-core includes a cooler in the box – no other CPU in the 5000 series lineup does.
So why would AMD raise prices? Well, first, the “duh” obvious – to make more money. At the high end, $50 more equates to a 10% MSRP bump on the 12-core CPU, which, while not inconsequential, isn’t that bad and hits at a market segment that can justify a 10% increase for a roughly 26% total average increase in performance. The percentage change is even lower at the 16-core level, and a relief as I thoroughly expected AMD to just go for broke and stick a $999 pricetag on it, since their existing 16-core already beats Intel’s 18-core HEDT part! It makes substantially less sense at the lower end of the stack, where the 8-core now firmly sits outside of the realm of possibility in many budget builds, as it both gains in price and loses value from not having an included cooler, meaning you’ll need to spend an extra $50 over the far less popular Ryzen 7 3800X just to get a foot in the door and then buy a CPU cooler on top of that. If you compare the x800x SKUs generation to generation, the increase is only 12.5%, but then factor in the cooler and if you were to instead compare the vastly more common 3700x to the 5800x, the picture looks much worse – a 36% price hike not accounting for the cooler!
Competitively, I fully understand where AMD is coming from. Being the budget brand and “value alternative” to Intel isn’t a good thing to company image and results, and if AMD is ever going to break the cycle of being neglected in mainstream system builds and mindshare, they need to offer a product that stands strong as the leader in all ways. This price hike accomplishes that by still allowing AMD the best performance per dollar title (because Intel pricing sucks) and allows them to look more premium while doing it. It is performative marketing bullshit, but it also works, and that is why we are here today.
For me, once I saw the performance and pricing of the 16-core? I was in – I expected far worse on price and lesser performance than AMD is pledging, and paying a small premium over last generation for huge gains makes a lot of sense for me. My problem is that at the low end, it is very tough to justify recommending Zen 3 to most of my friends and family. The 6-core would be the sweet spot for a lot of people, but at $300, it just doesn’t make sense. Now, I would fully expect that AMD could be playing a long game – there were no non-X SKU CPUs on offer in the new lineup as announced, and I fully expect that as third-generation inventory clears shelves, AMD themselves will offer discounts or official MSRP cuts later to the lower-tier parts to bring them more into line with expectations. Intel’s Rocket Lake launch in Q1 2021 should also facilitate this happening, but I think it is a bit of a mistake to so uniformly push prices up in $50 chunks.
If I were AMD, I feel like a $275 6-core would be a good compromise – a 10% increase over the 3600X starting price for a 26% performance uplift is a strong offer and one that would create better segmentation, in my opinion. Mentally, for me, there is a huge gap between $250 and $300, where my brain resists the increase to $275 far less. Intel’s i7-10700 offers 8 cores, integrated graphics, and a cooler at a current price per PC Part Picker of $310.99. With a higher boost clock and the two core advantage, it would make more sense in this price range to go with the Intel option, which is not something I’ve said in a long time! AMD’s bet here is a bold one – that a small but consistent lead in gaming performance on Intel is going to be enough to carry the pricing increase for the generation the increase is earned in. I’m skeptical that it’s going to be a good idea – with the carve-outs I’ve made above for the higher core-count parts being my exception, I think that it would have done AMD well to have launched this generation, owned on performance, and then increased prices. The way they are proceeding is fine if they expect Rocket Lake to be unable to wrestle back the gaming crown, but if Intel is back on top in the next 6 months, they’ll be forced to backtrack, and cutting prices as a competitive response nearly always reads as a loss in PR terms. AMD could be hedging by having non-X parts on deck for a competitive response in that event, but I couldn’t tell you how that would be perceived.
And so that leaves us here today – I still think the announcement is incredibly exciting, but I am also an AMD CPU fanboy to an extent and prefer their platform focus over Intel’s constant shifting of sockets and chipsets that stymies upgrading. It’s too early to say these are blunders on AMD’s part, because they could just as easily sweep the generation and leave Intel on their backs, and I think that seems like a likely outcome. The rumors about Rocket Lake don’t point to vastly higher performance and a lot of the mentions of the new core designs indicate that clock regressions are still pretty likely, so I would fully suspect that Intel is basically out of the competitive picture in many ways until they can get the “Cove” core designs running on the intended 10nm process on desktop with full power and higher clock speeds. However, AMD’s pricing does leave Intel an opening they can try to take to claim some of AMD’s mid-range stronghold, and I think that is a blind spot to AMD based on the news today.