5% lead in gaming is nothing to write home about, especially when it is achieved with fairly bog-standard B-die kits. The point is that there is still headroom left in Intel(with both core and memory OC), while there is none in AMD.
Just like overclocking and memory tuning is an outlier for the common gaming audience. They look at the stock gaming king and then they buy the SKU they can afford.
When Intel was in the lead in gaming, even 2% was "a big lead", not you are saying 5% is not a big lead ????
a 14 nm product is supposed to beat a 7 nm ?
But at stock AMD barely scrapes through, with losses in two games, and without the outliers I'd reckon that Intel and AMD are even in gaming. Affordability-wise Intel is clearly the better choice right now when comparing the 10/12 core parts.
Nah DOTA 2 is nothing like Doom Eternal. I think it barely scales with resolution unless you go all the way up to 4K.If the CPU can prepare and send more than enough draw calls to the GPU to keep it constantly busy, then the game is GPU bottlenecked. It's quite conceivable that a 10900K operating at roughly 5ghz would easily keep a 2080Ti GPU bottlenecked at 1080p high settings in Dota 2.
For comparison, a high FPS game like Doom Eternal at 1080p still received a large 40% performance boost when going from a 2080 Ti to a 3090, and on a 3900xt platform no less, indicating that the game was still GPU bottlenecked at 1080p and this game has one of the best implementations of Vulkan out there:
Custom-Designs der GeForce RTX 3090 im Test: Benchmarks in Full HD und WQHD (Update) - ComputerBase
For practical purposes I agree, but when performing a CPU performance test, the resolution should be set to 720p or less to minimize GPU interference.
I find your remarks to be in bad faith. We know very well that Zen 2 sees a massive boost with tuned RAM and same frequency. If latency is still an issue, then RAM tuning will still give good performance boost possibly negating the oc boost of intel.
The $50 rise for MSRP is unfortunate. But looking at other forums this was not an areas where AMD could have won over anybody skeptical anyway, most people were comparing street prices of Ryzen 3000 with these new MSRPs, of course the former are vastly lower with or without that imo rather small rise.
).I highly doubt they used the top card. Even in this video they first presented the 5900X to then top it with the 5950X later on. Giving away the best possible numbers as the tease for the next event would be out of character there.
So you're dismissing numbers at will, then go ahead and 'reckon' a mid-single digit lead equal, even though the 'outliers you're dismissing represent the most CPU-bound games.
Hm, somehow hoped for more details on the changes to the cores. Unified L3$ we knew for ages, and "Zero Bubble Branch Prediction" is the only new token brand we got to hear. What exactly are the improvements for Load/Store, and the changes to the issue width of int and float engines? I guess the event was squarely targeted at gamers, and have to hope wikichip or some other specialized site will get the info I'd like to hear about Zen 3.
The $50 rise for MSRP is unfortunate. But looking at other forums this was not an areas where AMD could have won over anybody skeptical anyway, most people were comparing street prices of Ryzen 3000 with these new MSRPs, of course the former are vastly lower with or without that imo rather small rise.
In AMD's case the MSRP always has been an early adopter premium. Honestly I consider the $50 rise rather low for what it is (and will wait for lower street prices as always
I highly doubt they used the top card. Even in this video they first presented the 5900X to then top it with the 5950X later on. Giving away the best possible numbers as the tease for the next event would be out of character there.
You mean the 2% they won in games ? They loose in everything else, productivity, power, etc... And the 5000 series just took away that, even though you won't admit ti. "Full deniability" I think is the term.....Like it does right now?
Show me tuned RAM results which show massive gains for Zen 2 that are NOT 720p benchmarks. You have 200 MHz headroom left on Zen 2 if you're overclocking memory and want to keep the DRAM:FCLK ratio at 1:1. On Intel you have a 800MHz headroom.
Do NIST recommendations include telling young gamers to ignore their favorite titles because their use case is not statically sound? What happens if this type of performance boost in Zen 3 gaming performance correlates with even more competitive titles which are built for a wider audience?
Do you even know how outliers are decided? Here have a read:
Facebook
Twitter