Backforce One Plus Gaming Chair review
ASUS GeForce RTX 3080 Noctua OC review
AMD Ryzen 5 5600 review
PowerColor RX 6650 XT Hellhound White review
FSP Hydro PTM Pro (1200W PSU) review
ASUS ROG Radeon RX 6750 XT STRIX review
AMD FidelityFX Super Resolution 2.0 - preview
Sapphire Radeon RX 6650 XT Nitro+ review
Sapphire Radeon RX 6950 XT Sapphire Nitro+ Pure review
Sapphire Radeon RX 6750 XT Nitro+ review
AIDA64: Update v6.00 brings Ryzen 3000 specific benchmarks
Yesterday we updated our Finalwire download with a v6.00 revision for AIDA64, which includes some major enhancements. Included are revised benchmarks and stability tests for AMD's upcoming Zen 2 processors, including Ryzen 3000 (Matisse) and Epyc 2 (Rome), which the chiplet design.
AIDA64 is used mostly used to read system data, but also comes with various benchmarks and stability tests. Useful is, for example, the memory test, which measures bandwidth and latencies of the CPU caches and the connected main memory. With the update to version v6.00, the provider Finalwire updates its test suite to make it fit for modern CPU microarchitectures. A new hashing benchmark can utilize, among other things, AVX-512 instructions.
Custom tests for AMD's Zen 2
In the run-up to AMD's release of the Ryzen 3000 processors, AIDA64 offers customized benchmarks and stability tests. Matisse, the codename for Zen 2 desktop procs, is highlighted. In addition, the new tools are also intended for Epyc 2 (Rome), but should be suitable for Ryzen Threadripper 3000. AMD's Zen-2 generation uses a new design in which the I/O area and memory controller are positioned onto a separate 14 nm die. The cores and their caches are in separate 7nm dies, the combinations are called chiplets.
New features & improvements
- SHA3-512 cryptographic hash benchmark utilizing AVX, AVX2 and AVX-512
- AVX2 and FMA accelerated 64-bit benchmarks for AMD Zen 2 Matisse processors
- Microsoft Windows 10 May 2019 Update support
- BeadaPanel LCD and SteelSeries Rival 710 OLED display support
- Corsair H100i Platinum and H115i Platinum liquid cooler sensor support
- Cooler Master MP750 RGB LED mousepad support
- Corsair Obsidian 1000D, EVGA iCX2, Farbwerk 360, NZXT GRID+ V3 sensor support
- Preliminary support for Intel Comet Lake CPU
- GPU details for AMD Radeon VII
- GPU details for nVIDIA GeForce GTX 1600 and GeForce RTX 2060 Series
- Preliminary support for AMD Navi GPUs
- Retired SHA1 and VP8 benchmarks
A new hashing benchmark supports a broad set of instructions: AVX, AVX2, AVX-512, BMI2, and XOP. AVX-512 controls Intel's Skylake SP and probably the complete Ice Lake family with Sunny Cove cores. AMD's Zen 2 should be limited to AVX2, which should not bother end users. Last but not least, Finalwire names the support of all current Vega-based graphics cards and Nvidia's complete Turing family.
You can download AIDA V6.00 here,
fantaskarsef
Senior Member
Posts: 12987
Joined: 2014-07-21
Senior Member
Posts: 12987
Joined: 2014-07-21
#5671784 Posted on: 05/22/2019 10:40 AM
I like that they seemed to include a stability test for Zen2 CPUs, but I am not sure if there's a need for a vendor specific benchmark... unable to be compared to Intel, AMD's competitor.
I like that they seemed to include a stability test for Zen2 CPUs, but I am not sure if there's a need for a vendor specific benchmark... unable to be compared to Intel, AMD's competitor.
asturur
Senior Member
Posts: 1202
Joined: 2010-05-12
Senior Member
Posts: 1202
Joined: 2010-05-12
#5671789 Posted on: 05/22/2019 10:47 AM
the benchmark could measure things that intel does not have. latency between chiplets or whatever.
It does not have to be power measurement or a speed index of some sort for all the cpu.
Also consider this is pretty normal considering not all cpus have same instruction sets, avx, avx2, sse4.1 and sse4.
the benchmark could measure things that intel does not have. latency between chiplets or whatever.
It does not have to be power measurement or a speed index of some sort for all the cpu.
Also consider this is pretty normal considering not all cpus have same instruction sets, avx, avx2, sse4.1 and sse4.
Astyanax
Senior Member
Posts: 13294
Joined: 2018-03-21
Senior Member
Posts: 13294
Joined: 2018-03-21
#5671835 Posted on: 05/22/2019 01:49 PM
Feiry does ASM level optimizations for each cpu, so in all fairness, aida's benchmarks are only valid between the same generation of cpu's.
I wonder if it means that developers have to utilize their software/engines for Zen2.
Feiry does ASM level optimizations for each cpu, so in all fairness, aida's benchmarks are only valid between the same generation of cpu's.
nevcairiel
Senior Member
Posts: 804
Joined: 2015-05-19
Senior Member
Posts: 804
Joined: 2015-05-19
#5671867 Posted on: 05/22/2019 02:57 PM
If you want ideal performance, thats pretty much true for every CPU - especially those that change their SIMD behavior. And in the case of Zen2, they are expanding their AVX/AVX2 SIMD to operate on full 256-bit, like Intel, instead of the 128-bit engine that Zen/Zen+ had.
I wonder if it means that developers have to utilize their software/engines for Zen2.
If you want ideal performance, thats pretty much true for every CPU - especially those that change their SIMD behavior. And in the case of Zen2, they are expanding their AVX/AVX2 SIMD to operate on full 256-bit, like Intel, instead of the 128-bit engine that Zen/Zen+ had.
Click here to post a comment for this news story on the message forum.
Posts: 6073
Joined: 2011-01-02
I wonder if it means that developers have to utilize their software/engines for Zen2.