AMD’s upcoming 12nm Ryzen refresh (known as Zen+) will be an interesting one, with the company touting improvements in clock speeds, soldered IHS and the inclusion of Precision Boost 2. In theory at least, this will help AMD fend off Intel’s Coffee Lake series of processors until AMD can put out the next generation Zen 2.
There’s been a plethora of information popping up on these processors, some by official channels along with the usual leaks and rumors. Back in mid January, 2018 there was a SiSoft Sandra entry which hinted at what we’d be seeing for this upcoming processor, but it was clearly an early engineering sample with less than final clock speeds. Still, it did by and large serve to confirm that core configuration of the Ryzen 5 2600 would remain essentially identical, from 6 cores and 12 threads, to the same amount of cache across the cores.
Fast forward to now, and we have another Ryzen 5 2600 leak, this time courtesy of GeekBench 4.0.3. The chip is identified as AMD Ryzen: ZD2600BBM68AF_38/34_Y and is running on the AMD Myrtle-PiR platform, complete with 16GB of DDR4 memory. Clock speeds have increased some 200 Mhz over both the boost and base clocks of the Ryzen 5 1600, with Ryzen 5 2600 clocked at 3.4GHz for the case, and boost clocks at 3.8 GHz. We still see 16MB L3 cache, and 3MB L2 cache (so that’s 512KB per core L2, and 8 MB x 2 of L3). From the looks of things, it’ll still be a 6W chip, essentially meaning the Ryzen 5 2600 will replace the 1600 in the same TDP bracket.
Of course, it’s hard to know whether the clock speeds (either of the base or boost) we’re seeing here are final retail sample clocks. In theory at least, the move from 14nm to 12nm should give AMD breathing room for about 10 percent improvement in clock speeds. But then AMD might decide to operate at lower temps and let users overclock. This would certainly improve yields, as it would mean that the clock speed goals that each chip had to reach would be lower. In essence, lowering the goal post of the quality of silicon and possibly allowing AMD to reduce costs of the chip. But this is all theory!
Enough of clock speeds, what about performance? Well here’s where things get interesting. The Ryzen 5 2600 scores 4,269 in Single Core and 20,102 in multi core performance with GeekBench 4.0.3. Those scores are certainly a bit better than you’d expect to find in a Ryzen 5 1600, so it’s hard to argue that the Ryzen 5 2600 is an improvement over its predecessor.
The problem is – there’s a lot of variables which can drastically alter performance and thus making it rather tricky to know how much of an improvement it truly is. Roughly speaking, the scores are about 10 percent improved in the single thread performance, and between 10 and 15 percent on multi core score. But – don’t forget we’re also looking at a clock speed advantage of 200 MHz for one.
Secondly, we don’t know RAM clock speeds or timings, which applications are running in the background (although highly unlikely), or other factors such as the cooling solution in place, whether these are the final clocks. There’s also a lot of rumors and hints that the Ryzen 2000 range will benefit from subtle tweaks to the cache, specifically tighter timings – which should help multi core scores. If a CPU on CCX 1 needs to retrieve info from a CPU core on CCX2, lower latencies will certainly improve things.
Ultimately, Pinnacle Ridge and Ryzen 2000 isn’t going to be a huge improvement – instead, it is a refresh. Think of it like Kaby Lake vs Skylake, subtle tweaks, improvements to PCIE bandwidth, better clock speeds. With that said, 10 – 15 percent improvement (assuming these numbers are similar across a large swathe of benchmarks in final retail silicone) isn’t anything to sneeze at. And if it overclocks fairly well to (say, 4.4GHz) Intel will have a compelling fight on its hands.