buildzoid posts
ASrock B650 LiveMixer PCB Breakdown notes
Vcore Renesas RAA229620 7phase14x Renesas ISL99360 60A SPSbulk output filter: 12x Nichicon FP12K 560uF TH poly bulk input filter: 4x Apaq 5K 270uF TH poly220nH inductors500KHz 1.2V SPS only200A 17W (typical overkill for AM5 boards)300A 28W400A 41W500A 57Wvoltage regulation and LLC:optimal LLC is level 3even at level 3 there's measurable undershoot because the load line is still very shallowthe board defaults to level 1 LLC everytime you change Vcore which is annoying.SOC Renesas RAA229620 +22x Renesas ISL99360 60A SPS?nH inductors2x Nichico...
Low effort Single rank Hynix 16Gb A/M-die timings for Ryzen 7000 at 6000Mbps
These timings should work on pretty much any hynix based DDR5 kit that is currently available(16Gb M-die and A-die at the time of writing). At DDR5-6000.VoltagesVDDIO / MVDD / MVDDQ: 1.35VVSOC: 1.25VTimingstCL: just use XMPtRCD: just use XMPtRP: just use XMPtRAS: 30*tRC: 68tWR: 48tREFI: 50000tRFC: 500tRFC2: 400tRFC4: 300tRTP: 12tRRD_L: 8 // Some hynix A-die really really sucks at tRRD_LtRRD_S: 4tFAW: 20tWTR_L: 16tWTR_S: 6tRDRDSCL: 4tRDRDSC: 1tRDRDSD: AUTO (only relevant for dual rank)tRDRDDD: AUTO (only relevant for multi rank)tWRWRSCL: 4tWRWRS...
RTX 4090 Zotac AMP Extreme AIRO PCB Breakdown notes
Dual BIOSInput filtering bulk capacitance32x 330uF TH polyVcoreController: UP9512 8 phaseMOSFETs: 24x AOZ5311NQI 55A DrMOS (3 per phase)1V 500KHz 150nH (power stages only?)300A ~24W400A ~36W500A ~48W600A ~63W800A ~115WVcore VRM bulk capacitance: 28x 1200uF TH polyVcore core bulk capacitance: 4x 470uF SMD poly2x 10x MLCCVmemController: UP9512 4 phasesMOSFETs: 3x AOZ5311NQI 55A DrMOS (1 per phase)Vmem bulk capacitance: 10x 330uF SMD polyNotes:This card probably has the highest total input and output filtering capacitance of any RTX 4090. Unfortun...
RTX 4080 Suprim X PCB breakdown notes
Fuses:12A on PCI-e slot20A for memory VRMsInput filtering bulk capacitance11x 180uF 16V TH aluminum polymer22x 100uF 16V SMD tantalum polymerVcoreController: UP9512R 8 phaseMOSFETs: 18x NCP302150 50A DrMOS (2 per phase and 3 in some)1.2V 500KHz 150nH (DrMOS only)300A ~27W400A ~38W500A ~57W600A ~78WVcore VRM bulk capacitance42x 220uF SMD aluminum polymerVcore core bulk capacitance2x 470uF SMD aluminum polymer3x 10x MLCCVmem Controller: UP9529Q 3 phasesMOSFETs: 3x NCP302150 50A DrMOS (1 per phase)Vmem bulk capacitance:9x 220uF SMD polyNOTEs:The c...
Gigabyte RTX 4090 Master PCB Breakdown Notes
Dual BIOSInput filtering bulk capacitance20x 100uF Tant SMD9x 270uF TH polyVcoreController: UP9512 8 phaseMOSFETs: 24x SiC653A 50A PS+OCP OTP HSSCP (3 per phase)1V 500KHz 150nH (SiC632 used for calculations)300A ~24W400A ~36W500A ~48W600A ~63W800A ~113WVcore VRM bulk capacitance:48x 470uF // that's A LOOOOT of SMD polymersVcore core bulk capacitance:4x 560uF SMD poly // usually you only 220-470uF here2x 10x MLCCVmemController: UP9512 4 phaseMOSFETs: 4x SiC653A 50A PS+OCP OTP HSSCP (1 per phase)Vmem bulk capacitance:10x 330uF SMD poly // sta...
PCB Breakdown notes: ASUS Strix B650E-F Gaming Wifi
Vcore ASUS ASP2208 6phaseVishay SiC659 80A SPS 2x per phase7x 6.3V 560uF TH poly 5K105C Apaq?7x 270uF 16V TH poly 5K105C Apaq?470nH inductors1.2V 500KHz (180nH) (based on SiC645 60A SPS)200A 17W 300A 28W (more than powerful enough for an overclocked 7950X)400A 47W500A 66WSOC ASUS ASP2208 +2phaseVishay SiC659 60A SPS x2 per phase3x 6.3V 560uF TH poly 5K105C Apaq?VDDMISCMaxLinear MXL 7630S 30A buck converter7x 6.3V 560uF TH poly 5K105C Apaq?1x 270uF 16V TH poly 5K105C Apaq?VDDIO_MEM richtek 6X=9B2x 4C10N Hi2x 4C10N LoFeatures2.5G lan intelPOST LE...
DDR5-6400 CL30 1T 2x16GB Hynix 16Gb M-die, Z790 Aorus Elite, i5 13600K
CPU: i5 13600K @5.5GHz P-cores, E-cores offCooling: fan on RAM, Asetek 240mm AIOmobo: Gigabyte Z790 Aorus EliteBIOS: F3fRAM: 2x16GB Corsair Vengeance DDR5 ES 6200C36 (5H16M) single rankRAM / IMC voltagesVCCSA: 1.35V // upto 1.45V is safeVCCIN: 2V // upto 2V is safeCPU VDDQ: 1.4V // upto 1.45V is safeCPU VDD2: 1.4V // upto 1.6V is safeRAM VDD: 1.45V // upto 1.6V is safeRAM VDDQ: 1.4V // upto 1.6V is safeRAM VPP: 1.8V // upto 2V is safePLL voltages:CPU core PLL: +120mvRING PLL: +120mvSA PLL: +120mvMC PLL: +90mvMemory training options:RTL training...
DDR5-7200 2x16GB Hynix 16Gb A-die, Z790 Aorus Elite, i5 13600K
CPU: i5 13600K @5.5GHz P-cores, E-cores offCooling: fan on RAM, Asetek 240mm AIOmobo: Gigabyte Z790 Aorus EliteRAM: DDR5 Gigabyte Aorus 7200C34 ES Single rank SK Hynix 16Gb A-dieRAM / IMC voltagesVCCSA: 1.35V // upto 1.45V is safeVCCIN: 2V // upto 2V is safeCPU VDDQ: 1.4V // upto 1.45V is safeCPU VDD2: 1.4V // upto 1.6V is safeRAM VDD: 1.4V // upto 1.6V is safeRAM VDDQ: 1.4V // upto 1.6V is safeRAM VPP: 1.9V // upto 2V is safePLL voltages:CPU core PLL: +120mvRING PLL: +120mvSA PLL: +120mvMC PLL: +90mvMemory training options:RTL training: Enable...
2x16GB Hynix M-die 6320 CL30 with a Ryzen 9 7950X on a Gigabyte X670E Aorus Master
EDIT: I set this up based on incorrect information provided to me by motherboard vendors. DO NOT USE THESE VOLTAGE SETTINGSCPU: Ryzen 9 7950XCooling: fan on RAM, EVGA CLC280 on CPUmobo: Gigabyte X670E Aorus MasterRAM: DDR5 Corsair Vengeance ES 2x16GB Single rank SK Hynix 16Gb M-dieRAM / IMC voltagesVSOC: 1.4V // EDIT: NOT SAFE VMISC: 1.1V // upto ??? is safeVDDG CCD: 950mv // upto ??? is safeVDDG IOD: 950mv // upto ??? is safeVDDIO_MEM: 1.38V // upto 1.4V is safeVDDP: 1150mv // upto ??? is safeRAM VDD: 1.43V // upto 1.6V is safeRAM VDDQ: 1.43V ...
DDR5-7000 2x16GB Hynix 16Gb M-die, Z790 Aorus Master, i9 13900K
CPU: i9 13900K @5.6GHz P-cores, E-cores offCooling: fan on RAM, EVGA CLC280 on CPUmobo: Gigabyte Z790 Aorus MasterRAM: DDR5 Corsair Vengeance ES 2x16GB Single rank SK Hynix 16Gb M-dieRAM / IMC voltagesVCCSA: 1.35V // upto 1.45V is safeVCCIN: 2V // upto 2V is safeCPU VDDQ: 1.3V // upto 1.45V is safeCPU VDD2: 1.45V // upto 1.6V is safeRAM VDD: 1.55V // upto 1.6V is safeRAM VDDQ: 1.5V // upto 1.6V is safeRAM VPP: 1.9V // upto 2V is safeRAM speed and timingsSpeed: 7000MbpsPRIMARY TIMINGS----------------------------------tCL: 34tRCD: 42tRP: 42tRAS: ...
4x16GB Hynix 16Gb M-die, Z790 Aorus Master, i9 13900K
CPU: i9 13900K @5.6GHz P-cores, E-cores offCooling: fan on RAM, EVGA CLC280 on CPUmobo: Gigabyte Z790 Aorus MasterRAM: DDR5 Corsair Vengeance ES 4x16GB Single rank SK Hynix 16Gb M-dieRAM / IMC voltagesVCCSA: 1.3V // upto 1.45V is safeVCCIN: 2V // upto 2V is safeCPU VDDQ: 1.3V // upto 1.45V is safeCPU VDD2: 1.4V // upto 1.6V is safeRAM VDD: 1.5V // upto 1.6V is safeRAM VDDQ: 1.45V // upto 1.6V is safeRAM speed and timingsSpeed: 6600MbpsPRIMARY TIMINGS----------------------------------tCL: 32tRCD: 40tRP: 40tRAS: 28tCR: 2TSECONDARY TIMINGS--------...
RTX 4090 Gamerock OC PCB breakdown notes
Dual BIOSInput filtering bulk capacitance21x 270uF TH polyVcoreController: UP9512MOSFETs: 16x NCP302150 50A DrMOS (2 per phase)1V 500KHz 150nH (power stages only?)300A ~28W400A ~42W500A ~62W600A ~91W640A ~104WVcore VRM bulk capacitance: 28x 820uF TH polyVcore core bulk capacitance: 4x 470uF SMD poly2x 10x MLCCVmemController: UP9512 3 phasesMOSFETs: 3x NCP302150 50A DrMOS (1 per phase)Vmem bulk capacitance: 9x 330uF SMD polyNotes:This is literally the worst RTX 4090 PCB I've seen so far.
RTX 4090 Suprim X PCB breakdown notes
Fuses!1x 12A fuse on PCI-e slot 12V1x 20A fuse for Vmem's 12V supplyInput filtering bulk capacitance13x 180uF 16V TH aluminum polymer28x 100uF 16V SMD tantalum polymerVcore Controller: MP2891 12 phasesMOSFETs: 26x MP86957 70A SPS (2 in most phases 3 in 2 phases)1.2V 500KHz 150nH (power stages only?)300A ~24W400A ~30W500A ~39W600A ~48W800A ~78WVcore VRM bulk capacitance48x 220uF SMD aluminum polymerVcore core bulk capacitance4x 470uF SMD aluminum polymer2x 10x MLCCVmem Controller: MP2891 +4 phasesMOSFETs: 4x MP86957 70A SPS (1 per phase)Vmem...
RTX 4090 Strix PCB breakdown notes
Input filtering bulk capacitance12x 270uF 16V TH aluminum polymer16x 56uF 16V SMD aluminum polymerVcore Controller: MP2891 12 phasesMOSFETs: 24x FDMF3170 70A SPS (2 per phase)1.2V 500KHz 150nH (power stages only scaled down from 1.8Vout)300A ~25W400A ~34W500A ~43W600A ~54W800A ~81WVcore VRM bulk capacitance14x 820uF 2.5V TH aluminum polymer35x 470uF 2V SMD aluminum polymerVcore core bulk capacitance4x 470uF SMD aluminum polymer2x 10x MLCCVmem Controller: MP2891 +4 phasesMOSFETs: 4x FDMF3170 70A SPS (1 per phase)Vmem bulk capacitance:18x 470uF S...
RTX 4090 Founders Edition PCB breakdown notes
Input filtering bulk capacitance13x 180uF 16V TH aluminum polymer13x 100uF 16V SMD tantalumn polymerVcoreController: MP2891 10 phasesMOSFETs: 20x MP86957 70A SPS (2 per phase)1.2V 500KHz 150nH (power stages only?)300A ~23W400A ~35W600A ~60W800A ~96WVcore VRM bulk capacitance: 40x 330uF SMD polyVcore core bulk capacitance: 6x 10x MLCC back of coreVmemController: MP2891 +3 phasesMOSFETs: 3x MP86957 70A SPS (1 per phase)Vmem bulk capacitance: 9x 220uF SMD polyNotes:I like it. High current handling capability thanks to the MP86957s. Lots of SMD fil...
Gigabyte X670E Aorus Xtreme Vcore regulation test results
All measurements are taken with a Rigol DS1054Z unlocked to 100MHz bandwidth.Ultra Extreme LLC is the absolute worst thing I've measured on an AM5 motherboard so far with 180mv of undershoot. It's so bad in fact that I was struggling to get a measurement before the CPU crashed due to the undershoot.Extreme LLC is absolutely atrocious with 150mv of undershootTurbo LLC is as bad as Extreme LLC with 150mv of undershootHigh LLC also sucks with 120mv of undershootMedium LLC also sucks with 110mv of undershootLow LLC also sucks with 100mv of ...
Gigabyte X670 Aorus Elite AX load line calibration test
All measurements are taken with a Rigol DS1054Z unlocked to 100MHz bandwidth.Extreme LLC is absolutely atrocious with 140mv of undershootTurbo LLC still sucks with 130mv of undershootHigh LLC also sucks with 120mv of undershootMedium LLC also sucks with 110mv of undershootLow LLC also sucks with 100mv of undershootStandard LLC is still bad with 90mv of undershoot (A good result would be around 20-50mv of undershoot depending on the platform)Normal LLC is the same as Standard LLCAuto LLC is the same as Standard LLCSince this testing was done wit...
ASUS Crosshair X670E Gene load line calibration test results
All measurements are taken with a Rigol DS1054Z unlocked to 100MHz bandwidth.Level 8 LLC is as expected absolutely horrendous with a worst case scenario undershoot of 140mv (difference between minimum measured voltage and average RMS)Level 7 LLC is better but still has a lot of undershoot at 90mvLevel 6 LLC is ok with around 60mv of undershootLevel 5 LLC really is the best. With just 40mv of undershootLevel 4 LLC doesn't do anything for the undershoot and somehow ends up adding a bunch of overshoot with maximums hitting around 1.37V instead...
X670 Aorus Elite AX PCB Breakdown notes
Vcore Infineon XDPE192C3B 8phaseInfineon TDA21472 70A SPS x2 per phase10x Apaq 5K 560uF TH poly 6x Apaq 5K 270uF TH poly400nH inductors400KHz 1.2V SPS only200A 16W300A 23W400A 32W500A 44WVSOC Infineon XDPE192C3B +2OnSemi NCP303160 60A SPS3x ??? 330uF SMD polyVDDMISCRenesas RAA229621 2phaseRenesas ISL99390 90A SPS2x Apaq 5K 560uF TH polyVDDIO_MEM Ritchtek RT8120D 1phase1x OnSemi 4C10N High2x OnSemi 4C10N LowCPUVDD18JoulWatt JW5068A 8A 1phase buck converterMemoryShielded Daisy chain8 layer PCBFeatures2.5G lan realtekPWR button RST buttonQflash+Cl...
X670E Aorus Xtreme PCB Breakdown notes
VcoreRenesas RAA229628 18phaseRenesas RAA220105 105A SPS10x Nichicon 10K 560uF TH poly + 4x Apaq 330uF SMD poly6x Nichicon 10K 270uF TH poly300nH inductors500KHz 1.2V SPS+L200A 15W300A 22W400A 32W500A 44WVSOCRenesas RAA229628 +2phaseRenesas ISL99390 90A SPS1x Nichicon 10K 560uF TH poly + 2x Apaq 330uF SMD poly300nH inductorsVDDMISCRenesas RAA229621 2phaseRenesas ISL99390 90A SPS2x Nichicon 10K 560uF TH poly300nH inductorsVDDIO_MEMRitchtek ??? 1phase1x OnSemi 4C10N High2x OnSemi 4C10N LowCPUVDD18JoulWatt JW5068A 8A 1phase buck converterFeatures1...
ASUS Maximus VII Gene PCB Breakdown summary
MATX6 Layer PCBVCCIN VRM --------------------------------------------------------------------ASP1257 8 phaseIR3535 drivers 200-1000KHzCDS87350 "upto 40A" dual n-fets220nH inductorsbulk output filtering: 7x 100uF 16V + 6x 560uF 6.3V 10K FPCAPsbulk input filtering: 6x 270uF 16V 10K FPCAPsVCCIN power losses500KHz 2Vout 5Vdrv mosfets only80A ~11W (160W)120A ~17W (240W going to the CPU!)200A ~35W (400W purely theoretical)VDDR VRM -----------------------------------------------------------------ASP1103 2phaseTI 87588N "upto 25A" dual ...
AHOC test equipment vs hardware purchasing decisions
So Rigol is now starting to sell these: https://www.rigolna.com/products/digital-oscilloscopes/hdo1000/The 100MHz bandwidth (intel docs recommend measuring Vcore regulation at 100MHz bandwidth) version is pretty much perfect for the voltage regulation measurements that I do since it has more than 1mv of resolution at the voltage levels I'm measuring and it has a HDMI output that I should be able to just feed into my capture card which would eliminate having to mess around with a webcam pointed at the screen of my "unlocked" Rigol DS...
4x8GB Hynix DJR on Ryzen
CPU: Ryzen 7 3700XCooling: Xigmatek Thor's Hammer + Thermaltake toughfan 12mobo: Gigabyte B550 Vision D-PBIOS: F15aRAM: Gigabyte Aorus RGB Memory 4400C19 + 4800C19 single rank Hynix 8Gb DJR!! DJR is very temperature sensitive without a fan even 3800 18-22-22 was erroring !!RAM / IMC voltagesVSOC: 1.15V // upto 1.2V is safeCLDO VDDP: 0.9V // upto 1.1V is safeVDDG IOD: 0.95V // upto 1.15V is safeVDDG CCD: 0.95V // upto 1.15V is safeVDDR: 1.6V // max safe voltage unknowDDR VTT: 0.8V // max safe voltage unknownRAM speed and timingsSpeed: 3800Mb...
2x8 Nanya 8Gb B-die on Gigabyte B550 Vision D-P
CPU: Ryzen 9 5950XCooling: Xigmatek Thor's Hammer + Thermaltake toughfan 12mobo: Gigabyte B550 Vision D-PBIOS: F15aRAM: Corsair Vengeance LPX 3000C15 2x8GB single rank Nanya 8Gb B-dieRAM / IMC voltagesVSOC: 1.1V // upto 1.2V is safeCLDO VDDP: 0.9V // upto 1.1V is safeVDDG IOD: 1V // upto 1.15V is safe VDDG CCD: 1V // upto 1.15V is safeVDDR: 1.46V // max safe voltage unknowDDR VTT: 0.73V // max safe voltage unknownRAM speed and timingsSpeed: 3800MbpsPRIMARY TIMINGS----------------------------------tCL: 15tRCDRD: 18tRCDWR: 8tRP: 14tRAS: 21tRC...
AHOC discord link
paddingpaddingpaddinghere it is: https://discord.gg/jv3xjxv
Does setting tRCDWR to 8 do anything?
Yes it does.Py Prime 2B running on a 4750G with 1xCh 1Rx8 Hynix 8Gb CJR2400Mbps (timings have more impact at lower memory clocks as each tick is longer)12-16/8-16-28-1T 23.862s 23.833s 23.749s12-16/16-16-28-1T 24.417s 24.513s 24.372s12-16/26-16-28-1T 25.715s 25.701s 25.712sOn many many(probably most) DDR4 memory chips it's possible to set tRCDWR to 8 with no effect on stability. As you can see from the Py Prime results this does lead to a slight performance increase. So there's 0 reason not set your tRCDWR to 8(if it's passing stres...
Samsung DDR4 8Gb B-die timing guidelines for Ryzen
Viable for 3600-4000Mbps but skewed towards 3800Mbps with 1.4-1.65V depending on RAM chip quality. With GDM enabled.Note that most Ryzen 3000/5000 CPUs will not do more than 1900FCLK stable.Also Keep in mind that B-die is the least consistent DDR4 IC that I'm aware of. Unbinned/OEM B-die and low XMP spec B-die like 3200 16-18... can often struggle to scale to even 1.45V while being extremely temperature sensitive and doing horrific tCL tRCDRD and tRP. Like 3600 18-18-18 at 1.4V levels of horrific. On the flip side it is still B-die and so s...
Simplified RAM timing "definitions" WIP
RAM timings are how much time the RAM is given to complete various commands/operations. How quickly the RAM can complete a command/operation depends on the circuitry of the memory chip, the voltage it's running on and the temperature it's at. If you don't give the RAM chip enough time to complete a command you end up with missing/corrupted data. When you're lowering a timing you are merely trying to find the shortest amount of time in which the RAM chip can reliably complete whatever command/operation that timing governs. For ex...




