I wish i could share fine FPS values for GTA5, but I had always played with borderless windows, so that the FPS were always 31-32. Now, in fullscreen with new APU and OC, I get 60FPS. Good RAM in Desk- or Gamemini are very important.
For GTA V, remember to deactivate the "MSAA" option in advanced graphics settings, this option is much too heavy to manage for an integrated graphics card with only 8 ROPs and 32 TMUs, deactivating this option is saving for this iGPU is day and night, in 1080p (Normal quality with some settings on "high", the sliders are left stock) I obtain an average between 85 and 90fps over several hours of play depending on the game, be careful I have a "pocket PC" configuration with a Ryzen 7 5700G supercharged as much as I could: in the following config:
Case: J-HACK Lander 55 (Open duct to 12cm principal CPU fan)
APU: AMD Ryzen 7 5700G (CorePerformanceBoost On - iGPU Vega8 @2400MHz vSOC/vGFX@1,25v rockstable)
CPU Heatsink: ID-Cooling IS-50X (rev V3 - 400g - 5×Ф6mm Heatpipe Direct Touch v4.0 technology) modded with SilverStone Air Slimmer 120 ARGB(120mm(W)x120mm(D)x15mm(H) and very good performance for a 12cm ARGB Slim Fan : Max airflow of 64cfm et max static pressure of 2,4mmH²O and all in a max very relative noise rate at 31,7 dBA, this heatsink/fan assembly must have a capacity to support a maximum TDP of 150 Watts mounted with excellent Thermalright TF9 thermal paste (14W/m-k)
Motherboard: ASRock X300M-STX (rev1.02) from ASRock DeskMini X300, The BIOS is flashed in the latest “beta” version available “L1.80F”
Mod: Stock "plastic" thermalpad were removed to be replaced by a custom-cut thermalpad from Thermalright Valor Odin (15W/m-k)
RAM: Corsair Vengeance SO-DIMM 32GB(2x16) 3000MHz CL16-18-18-39 1.2v(value ram... yes but with only Samsung B-Die chips on it)
OC@4000MHz CL16-18-18-18-36-54-1T 1.35v with MemoryContreller(InfinityFabric) which synchronized with it(IF 1:1), the sticks were modded by me and had their "paper" labels removed, then were cleaned of all traces of glue or other, to have a passive dissipator applied on both sides entirely in C101 copper of 1.5mm thickness each, covering and protecting the entirety of both sides of their PCBs, and these heatsinks being entirely "painted" or "tempered" in a graphene solution (one of the best thermal conductors in the world) so that by this "light" coating graphene, the heatsinks evacuate their surface thermal releases a little better (between 20 and 30% am, according to the few "conclusive" and reliable tests that I have been able to find).
And in hindsight, it is above all this overclocking which offers the biggest gain in noticeable performance in fps in games and other 3D applications, between having a DDR4 32GB (2x16) kit at 3200MHz CL20 and my kit at 4000MHz CL16 well adjusted, I would say the gains vary from 30 to 50% more fps
Case: J-HACK Lander 55 (Open duct to 12cm principal CPU fan)
APU: AMD Ryzen 7 5700G (CorePerformanceBoost On - iGPU Vega8 @2400MHz vSOC/vGFX@1,25v rockstable)
CPU Heatsink: ID-Cooling IS-50X (rev V3 - 400g - 5×Ф6mm Heatpipe Direct Touch v4.0 technology) modded with SilverStone Air Slimmer 120 ARGB(120mm(W)x120mm(D)x15mm(H) and very good performance for a 12cm ARGB Slim Fan : Max airflow of 64cfm et max static pressure of 2,4mmH²O and all in a max very relative noise rate at 31,7 dBA, this heatsink/fan assembly must have a capacity to support a maximum TDP of 150 Watts mounted with excellent Thermalright TF9 thermal paste (14W/m-k)
Motherboard: ASRock X300M-STX (rev1.02) from ASRock DeskMini X300, The BIOS is flashed in the latest “beta” version available “L1.80F”
Mod: Stock "plastic" thermalpad were removed to be replaced by a custom-cut thermalpad from Thermalright Valor Odin (15W/m-k)
RAM: Corsair Vengeance SO-DIMM 32GB(2x16) 3000MHz CL16-18-18-39 1.2v(value ram... yes but with only Samsung B-Die chips on it)
OC@4000MHz CL16-18-18-18-36-54-1T 1.35v with MemoryContreller(InfinityFabric) which synchronized with it(IF 1:1), the sticks were modded by me and had their "paper" labels removed, then were cleaned of all traces of glue or other, to have a passive dissipator applied on both sides entirely in C101 copper of 1.5mm thickness each, covering and protecting the entirety of both sides of their PCBs, and these heatsinks being entirely "painted" or "tempered" in a graphene solution (one of the best thermal conductors in the world) so that by this "light" coating graphene, the heatsinks evacuate their surface thermal releases a little better (between 20 and 30% am, according to the few "conclusive" and reliable tests that I have been able to find).
And in hindsight, it is above all this overclocking which offers the biggest gain in noticeable performance in fps in games and other 3D applications, between having a DDR4 32GB (2x16) kit at 3200MHz CL20 and my kit at 4000MHz CL16 well adjusted, I would say the gains vary from 30 to 50% more fps
SSD: Crucial P5 Plus 2TB PCIe 4.0 x4(Read:6600MBps/Write:5000MBps) but the current APU only supports PCIexpress 3.0 so in PCIe Gen3 "mode" result(approx.) in 3550MBps for Linear Read an 3350MBps for Linear Write
SSD Heatsink: Thermalright HR-09(height 47-48mm) with thermalpad Thermalright Extreme Odyssey II(14,8W/m-k)
External PSU: Delta Electronics ADP-180HB D 180 Watts(19V 9.5A) 5.5x2.5mm - The 120 Watts provided by the AcBel 120W external power supply are no longer sufficient with this APU which has the CorePerformanceBoost activated, the iGPU and the IMC (InfinyFabric) overvolted and overclocked, the consumption at full load can reach 130 to 140 Watts, I switched to 180 Watts to avoid the problems of a burnt out power supply in the medium to short term, and above all so that the APU can use as much energy as soon as it needs it and so far on the "monitored" electrical outlets, this reached a peak of 167.8 Watts of electrical consumption during a long period of x265 encoding.
Last edited: