Possibles fuites sur la performance des nouveaux processeurs de génération prochaine.
Possibles fuites sur la performance des nouveaux processeurs de génération prochaine.
To clarify upfront, this piece is written in Persian, which I can't understand, and I'm unsure about the reliability of these figures. So I think it doesn't fit into tech news. Moving on to the story: the link points to an article at Sakhtafzarmag.com discussing benchmark results for Intel's Alder Lake CPUs. A tweet from @momomo_us shared a chart comparing performance against top AMD 5000 series CPUs, noting better or comparable speeds at lower prices. In my view, if this were accurate, AMD would need to adjust pricing to align with the reported performance, which would likely require a public announcement and actual product availability. Edit: The embedded tweet didn't display correctly; made changes on September 10, 2021 by StephanTW.
I believe AMD may reduce desktop prices if Alder Lake delivers stronger performance. They might also release the Threadripper 5000 series alongside Alder Lake to shift some attention away from it at the premium end. The 12900K could still feel like a weaker version of the Celeron when compared to the 5990X. I hope Intel responds by introducing a 24-core 12980XE to match the low-end Threadripper CPUs. I’d prefer AMD not to become too comfortable in the high-end HEDT market.
I notice the same 10% figure repeatedly appearing in single-core settings—it's not ideal, but it's a concern. Multi-core performance seems comparable. The standout here is the DDR5 4800, which I need to evaluate more closely at higher speeds. Despite the excitement around it, leaks have been quite disappointing so far, and I'm leaning toward Raptor Lake right now.
I believe we’re on the verge of a significant performance boost within the next couple of years. Alder Lake, Raptor Lake won’t satisfy my needs if I upgrade from an 8700k (gaming point of view). Looking at transistor densities makes it clear: - AMD using TSMC 7nm offers 96.5 MTr/mm² (Zen3/3dv) - Intel’s 10nm will provide 100.76 MTr/mm² (combined with Alder and Raptor Lake?) - TSMC’s 5nm EUV is projected at 171.3 MTr/mm² (Zen4) - Intel’s 7nm is estimated at 237.18 MTr/mm² (Meteor Lake) - TSMC’s 3nm is around 292.21 MTr/mm² - IBM’s 2nm is estimated at 333.33 MTr/mm² I think DDR5 will definitely improve, but I also expect CPU performance to rise so quickly that choosing the latest every time might feel a bit frustrating. This situation forces me to carefully consider, speculate wisely, and make smart investments based on performance, price, and timing. A CPU upgrade should ideally deliver the biggest gains possible without causing buyer’s remorse. A five-year lifespan is expected to be both practical and worthwhile. The challenge remains: if I pick Alder Lake, the boards might not be ideal and DDR5 memory speeds could lag behind Raptor Lake, which would only be a year later. They’ll also likely increase the L3 cache size on the i9 for better efficiency, and expect at least 200MHz faster RAM. Overall, there’s a clear maturation path that will likely move quickly, as Intel continuously pushes boundaries to regain leadership. There are plenty of examples—Zen1, Zen2, Zen3 all needed time to reach parity. The 8700k example stands out because it hasn’t delivered enough meaningful gains for gaming yet. Sure, there are improvements in RAM speeds and boards, but they haven’t translated into noticeable performance jumps. Besides a bigger L3 cache with the 10900k, the question is whether Intel’s CPUs will keep improving dramatically or just offer minor upgrades. What I’m also considering is: from 2023 to Meteor Lake, then Nivdia 4000 series in 2023, to 480Hz monitors—so technically, 2023 marks a major leap. It seems likely that a real performance boost is on the horizon within the next couple of years. Every day I think about it, I’m leaning toward waiting another two years and sticking with my current setup. The more I reflect, the more I believe yes… I might be able to wait and see what comes next. All I’m concerned about is saving money while still enjoying solid performance.
There are still delays in software updates. Even with significant improvements in 2023, most titles will remain constrained by the lower-end Zen2/RDNA found on upcoming consoles for the next five years. I might end up being a bit caught off guard, but it probably won’t really affect how much you play.
Hope we get some leaked benchmarks soon too. How strong would it be compared to the 3000 series? Would it fit in the same price range as the 3000 series?
Assuming a similar boost for the 5990X versus the 3990X compared to the 5950X versus the 3950X and 5900X versus the 3900X, we can expect roughly a 10-20% boost in multi-core performance. This would place the Cinebench R23 score between 81,800 and 89,300. The 5950X currently sits around 28,600. If the 12900K actually improves by about 10%, as suggested by the R20 numbers in a late July leak, it could reach approximately 31,500. However, I suspect the leak might have shown an overclocked version of the 12900K rather than the standard model. Intel boards sometimes enable features like MCE by default or adjust power limits, which technically counts as an overclock even without manual BIOS changes. If this is true, the 5950X may still edge out, especially when pushed to 4.7GHz where it gains around 20% in Cinebench R23 (about 34,300). All these points remain guesses until official testing is released. We’ll find out soon enough how this turns out.
I lack direct knowledge, yet I think AMD is likely increasing Threadripper inventory and preparing a launch either soon before or soon after the 12th generation. Both generate significant interest among journalists and fans, but I believe they’ll do it once comparisons are ready—so that they can highlight the full lineup versus Intel’s offerings. They’d need visuals demonstrating how the entire Threadripper range outperforms Intel’s top models, which will only happen after the Intel embargo lifts.
The 3960X was priced at $1,399. If the 5960X matches that cost, a $600 Intel chip delivering roughly 75% of that performance could still seem appealing unless you require quad-channel memory or many PCIe lanes. Sadly, I think the 5000-series Threadripper will follow the desktop trend: raising prices across the board. While the hype around Threadripper might spark interest, it won’t extinguish it entirely, especially once Intel secures a competitive edge.
My concern is that AMD may lower prices for the 5000 series components. If the 5950X matched the 12900K’s price, it could still be a strong option thanks to its more established design and likely cheaper boards. However, Intel’s support for DDR5 and PCIe 5 makes its platform appealing despite early limitations. At price parity, choosing between them would be tough.
If the 12900K becomes competitive, the 5950X might seem less compelling. Yet, I doubt AMD will match that, since their strategy seems focused on maintaining value rather than undercutting.
In my view, AMD’s best move could be launching budget-friendly desktop parts—like a $220 5600 non-X, $160 5300X, and $120 5100—especially if the 12400 stays at the lowest tier with a modest price jump over the 11400. Putting a 5100 on an affordable B550 board for under $200 could really shake Intel’s position in the budget segment.
Lowering prices for the 5300G for DIY use or reducing costs on the 5600G and 5700G would also help. The challenge remains: AMD has little motivation to reduce sales unless Alder Lake gains significant traction.