SoylentNews is people
SoylentNews
Intel's Alder Lake big.LITTLE CPU design, tested: It's a barn burner:
After spending several days with Intel's newest consumer CPU designs, we have some surprising news: they're faster than AMD's latest Ryzens on both single-threaded and most multithreaded benchmarks.
We suspect this will be especially surprising to some, since Intel's newest desktop CPUs feature a hybrid "big.little[sic]" design similar to those found in ARM CPUs. AMD's flagship Ryzen 9 5950x is a traditional 16 core, 32 thread design, with all cores being "big" high-performance types with symmetric multithreading (SMT, also known as "hyperthreading"). By contrast, the i9-12900K offers 16 cores and only 24 threads—with eight "performance" cores featuring SMT and eight lower-performance "efficiency" cores with no SMT.
As pointed out in the Ars Technica comments, the Cinebench multi-threaded benchmark saw Intel's best CPU with a less than 2.5% lead, but the caption reads "Intel trounces AMD". While the Passmark multi-threaded benchmark saw AMD's best CPU with a more than 18% lead, but the caption reads "outperform i9-12900k-but even here, by a much, much, lower margin than we're accustomed to seeing".
Also at Phoronix, AnandTech, and Tom's Hardware.
See also: More Linux Performance Benchmark Data For Alder Lake, Comparison Data Points
Intel UHD Graphics 770 / Alder Lake GT1 Linux Graphics Performance
Previously: Intel Alder Lake CPUs Launch November 4th, with Up to 8 Big and 8 Small Cores
(Score: 0) by Anonymous Coward on Friday November 05 2021, @09:07AM (1 child)
There's a reason why more people are interested in Ryzen than they were in Bulldozer. Basically at one point AMD had inferior chips. All they could think of was to increase the core count. But that's just stupid, because if I had lots of processing tasks that's helped merely by increasing core count I'd run that stuff on a GPU (well that was back when GPUs were more affordable ;) ).
In contrast the Ryzen stuff is genuinely fast, especially the newer stuff.
There were other good points of using AMD CPUs - many supported ECC and AMD tended to not change the sockets as often as Intel. But the latter isn't really a big deal for most people. I'm more likely to buy a entire new PC after X years rather than just change the CPU (especially if it was the old days when you weren't going to get much of a speed increase by upgrading to a newer AMD CPU - you'd be better off spending the upgrade money on SSD, GPU, or RAM).
(Score: 2) by Marand on Friday November 05 2021, @09:30PM
The real mistake with Bulldozer was the "cores" were integer cores with shared floating point units, so an "8 core" bulldozer only had 4 FPUs shared between core pairs. Both companies had their own idea for how to "fake" extra cores: Intel did SMT with "hyperthreading", while AMD tried a hybrid approach with more "real cores" that were only real for certain kinds of workloads. CPUs get designed years in advance and AMD made a bad prediction, apparently expecting integer performance to be king, when that didn't happen at all.
I think JavaScript may be partly to blame there since it doesn't have a real integer type. Bulldozer was being designed pre-Web2.0 when JS was a curiosity that just added minor features to websites, but was released into a world there JS-heavy sites were quickly becoming the norm and writing "native applications" with Electron quickly (and unfortunately) became popular. That put an unexpectedly high emphasis on floating point even in workloads that weren't as reliant on FP performance.
AMD gambled and lost, and nearly went out of business from it. With Ryzen they abandoned the failed experiment, adopted SMT like Intel did, and made other improvements, and it saved them.