As the days are passing, the long-awaited reveal of new Ryzen CPUs draws near. If you are, by any chance, following the hardware industry you might be aware that AMD has been crushing Intel for the last few years. The mainstream release of Zen 2 with the Ryzen 3000 chips started to turn the tide of market domination towards AMD as they offered much faster processors for drastically lower prices. This trend continued onwards with the release of Zen 3 and the Ryzen 5000 chips. Nevertheless, this release also increased the MSRP of the processors, albeit substantially reinforcing the superior AMD architecture as opposed to measly Intel 11th Gen CPUs. However, Alder Lake, the 12th generation Intel CPUs absolutely destroyed the Ryzen 5000 series with the Intel i9-12900k coming on top in almost all benchmarks possible. Around this time, was when I built myself my new computer. An Intel i9-12900K, 32GB RAM, RTX 3080 and so on. Now, only a couple of days to go for the AMD reveal, we are slowly getting the leaks in terms of "what to expect" from both sides. According to VideoCardz (and their sources) i9-13900, with just 65W TDP is able to deliver 2.13k in single-threaded Geekbench and 20.1k on multi-threaded Geekbench test. You can see the comparison between the 13th Gen, 12th Gen and AMD's 5000 series in the chart above. Greymon, a renowned leaker for AMD products has teased these: https://twitter.com/greymon55/status/1562831148904230912 To decipher, Ryzen 5 7600 and Ryzen 7 7700 are getting 19xx and 20xx single threaded and 198xx and 151xx multi threaded scores respectively. When we consider the AMD advantage in terms of nanometer, the results were more diminishing than I initially assumed. It does not seem like there's data for the 7950X, which would be the main competitor of the i9-13900k yet, but we can more or less assume that even 100 pts increase due to the silicon lottery and binning of high-end chips such as 7950X, around 2.1k single-threaded performance does not really come close to Intel. But does that mean Intel wins? Not really. There are advantages to Intel architecture just like there are advantages to AMD architecture. Not every workload is the same and AMD has been well-known for more "computing" related workloads with the efficiency of their humongous L3 Caches as well as certain features like AVX-512 not being fused off post-production. I am looking forward to the official AMD reveal on the 29th. Who knows, if we are talking +30% performance increase in both single thread and multi threaded work loads (be it AMD or Intel) I might just upgrade. Now... here's the more "tasty" part. In case you missed it, Intel ARC GPUs have been a failure and they have outright ceased production on this lineup. In the past two generations since the 2000 series for NVIDIA and 5000 series for AMD GPUs, there have been a major increase in the power usage for GPUs. The RTX 3000 series pushed this even further with RTX 3090 taking upwards of 400W in peak use. Does anyone remember the days when 250W GPUs were memed because they used a lot of power? Well, something even more sinister is on the way. First, lets see how NVIDIA is coming for the next gen GPUs: https://twitter.com/QbitLeaks/status/1563248106468438016
Source
Source
https://twitter.com/QbitLeaks/status/1551882582303813632
- QbitLeaks
According to the leaker, while RTX 4090 is more or less set in stone, prior to the launch in January, 4060, 4070 and 4080 SKUs are still a work in progress and is being altered every other week (according to some production leaks) with different SKUs popping up here and there. So what is this TSE that the leaker has posted and got everyone so surprised?
TSE stands for "Time Spy Extreme" it is a major, well-known benchmark used for GPU testing. According to QBitLeaks, the 4090 is expected to receive about 20.1k, 4080 ~13.4k and 4070 and 4060 seems to have received ~8.6k and 6k respectively.
For the record, the RTX 3090, in the hands of overclockers (with use/modifications that prevent daily driving and is only for OC purposes) get around ~13k. RTX 3080 gets around 8.9k -- in other words, if these leaks were to be true, it would mean that a RTX 4060 Ti delivering around the same performance as an RTX 3080.
Crazy!
Unfortunately, for AMD's side of things, things are not looking "that" good. For many years at this point AMD has been trying to catch up to NVIDIA in the GPU market, however, in most cases there's still five to ten percent performance difference depending on the task. According to some of the leaks, the spec sheet of the AMD cards look relatively underwhelming. Especially in terms of memory bandwidth with the utilization of GDDR6 instead of GDDR6X chips, therefore resulting in slower overall bandwidth.
I don't really have too much of an opinion for AMD though, I was not able to find any tangible leaks especially in regards to performance other than the spec sheet. So I am excitedly waiting for the 29th event. I am assuming they would be revealing the GPUs alongside the CPUs, because... why not?
Well, that was a long write-up. I just love hardware. I don't know if anyone else besides @themarkymark does. But if you do, comment your opinions!