Crosshair VII Hero Essential Info Thread

[gupsterg]

I ran 12 instances of HCI with hwInfo running. I didn't run long as I don't have much time on tonight. The allcore speeds were higher than normal. Generally all core speed is a lousy 3950Mhz. Should be better tomorrow so will repost then when I have sorted my cooling etc. This is at stock with no PBO + 3600Mhz ram @ 1.4v SoC @ 1.05v cLDO VDDG @ 0.95v

Attachment 81621

To me it would seem your CPU sample is on totally differing voltage/frequency curve than mine , due to it's unique silicon characteristics.

See my stock CPU OC RAM and PBO+150MHz OC RAM WMVs in this ZIP. Compare average CPU frequency and average CPU Core Voltage SVI2 TFN.

I feel I need to clarify that it's not exactly that I have no AMD CBS options, but only a severely reduced set.
I've been wondering if that's just how the new bioses are or if it's the case that a gen 1 ryzen are locked out of most options for whatever reason.
All I can acces is in the pic. Attachment 81625

If the options have been suppressed and not removed I maybe able to bring them back. I will check if I have time today.

Just wanted to chime in with another "thank you" for this, it's nice to have these custom combinations of BIOS+FW available as the older AGESA certainly makes a difference.

Just flashed your modded 0002 on a C7H+3700X and spent a while messing around, some browsing plus a CB20 single+multicore run. I actually seem to be getting even better boosting than I previously did on 2406/2501. 1 core @ 4.475GHz, 1 @ 4.45GHz, 1 @ 4.425GHz, 2 @ 4.40GHz, 3 @ 4.375GHz. When I used 2406/2501 I never saw any core go beyond 4.425GHz. CB20 multicore is up by ~100 points vs 2606 as well.

Definite thumbs up from me!

No problem, thank you for feedback and glad to read it shows gains .

Just as a share to anyone wondering. This ZIP has ~7hrs P95 4K 4096K 27000MB non AVX run, frequency/voltage seems as it would be on unmodified AGESA 1.0.0.2/1.0.0.3 UEFI. So the SMU FW mod is not making CPU unsafe in high load/current situation from what I have seen.

[mikev190]

To me it would seem your CPU sample is on totally differing voltage/frequency curve than mine , due to it's unique silicon characteristics.

See my stock CPU OC RAM and PBO+150MHz OC RAM WMVs in this ZIP. Compare average CPU frequency and average CPU Core Voltage SVI2 TFN.

Your voltage is much lower! I wonder if there is anything I can do to lower it. I mean manually I can run 4.1GHz @ 1.256v so it feels like it is putting way too much voltage in for the clock speeds. I've just sorted my cooler so I will do some more tinkering! I've noticed when the CPU is around 30c it will boost to 4.2GHz but with that voltage it is almost impossible to keep it that cool without water. Doing offset voltage lowers the clock speed

Hwinfo new beta has added a CCD temp I've just noticed


HWiNFO v6.11-3900 Beta available.

Changes:

Fixed reporting of ES/Production stage for some Zen2 CPUs.
Fixed measuring of BCLK for some Intel CPUs under Windows 7.
Added reporting of CPU High Temperature Clock Limit for AMD Zen2 family.
Added reporting of CPU Automatic Overclocking Offset for AMD Matisse family.
Added a new (more reliable) method for measuring BCLK on AMD Matisse systems.
Added monitoring of Memory Controller Clock (UCLK) for AMD Zen family.
Added preliminary support of Intel Elkhart Lake.
Fixed reporting of NVMe capacity when LBA data size > 512B.
Added reporting of CPU Thermal Trip Limit and HTC Temperature Limit for AMD Zen.
Updated reporting of CPU VDD and SoC SVI2 current/power on some MSI mainboards.
Added monitoring of per-CCD Tdie temperatures for AMD Zen2.

https://www.hwinfo.com/download.php

[gupsterg]

Your voltage is much lower! I wonder if there is anything I can do to lower it. I mean manually I can run 4.1GHz @ 1.256v so it feels like it is putting way too much voltage in for the clock speeds. I've just sorted my cooler so I will do some more tinkering! I've noticed when the CPU is around 30c it will boost to 4.2GHz but with that voltage it is almost impossible to keep it that cool without water. Doing offset voltage lowers the clock speed

Hwinfo new beta has added a CCD temp I've just noticed

If you offset it too much you may experience clock stretching. You may see same/higher MHz but not the performance. Again depends on your CPU.

From what you state "Doing offset voltage lowers the clock speed", IMO the SMU is lowering the clocks, due to how it profiled CPU, it thinks stability would be compromised, if clocks remained higher with lowered voltage.

I think there is little you can do. I think you have 3 choices:-

i) Live with CPU as is.
ii) Manual OC.
iii) Get another CPU.

[mikev190]

If you offset it too much you may experience clock stretching. You may see same/higher MHz but not the performance. Again depends on your CPU.

From what you state "Doing offset voltage lowers the clock speed", IMO the SMU is lowering the clocks, due to how it profiled CPU, it thinks stability would be compromised, if clocks remained higher with lowered voltage.

I think there is little you can do. I think you have 3 choices:-

i) Live with CPU as is.
ii) Manual OC.
iii) Get another CPU.

Yeah its a bit of a stinker. On the latest AMD SMU I don't ever hit 4.2Ghz so feel a bit ripped off tbh (thankfully you modded the bios, but thats our secret :P)
I can use the BCLK to get an extra 100MHz so might have to do that for now.

[gupsterg]

Yeah its a bit of a stinker. On the latest AMD SMU I don't ever hit 4.2Ghz so feel a bit ripped off tbh (thankfully you modded the bios, but thats our secret :P)
I can use the BCLK to get an extra 100MHz so might have to do that for now.

I have used BCLK in the past to gain extra on 2700X. Below is PE: Default PBO: Enabled, I didn't use PE1/2/3/4, were too aggressive on voltage IIRC, so using BCLK was best balance.

[mikev190]

I have used BCLK in the past to gain extra on 2700X. Below is PE: Default PBO: Enabled, I didn't use PE1/2/3/4, were too aggressive on voltage IIRC, so using BCLK was best balance.

Ah cool good to know! Thanks! I've been playing with manual OC just to get an idea of what clocks to voltage ratio it does. I don't want to manual OC due to lack of downclocking/volting? Or can you do it with a P-State like the 1700X? When I tried P-State it wouldn't budge the clock speeds with Minimum processor set to 5%.

I do feel like the CPU should be doing around 4150MHz in games instead of 4050MHz as with manual OC I can push 4150Mhz at 1.325v so don't understand why it doesn't do that with precision boost.

Anyways I think with BCLK that will be a workaround for now. I don't think it would ever boost as great as yours, but I do think it could be doing a little better. Perhaps future SMU/AGESA revisions might tweak Zen 2 boosting some more. One can hope anyways lol. When I do replace I think I will go for the 3700X or next-gen depending how long it takes me to cave in and buy another lol.

Again thanks for all the help and putting up with my rambling! I live in a small town and bar my dad there is no one to have tech/hardware discussions lol. So this is my only avenue for said conversations

[gupsterg]

Ah cool good to know! Thanks! I've been playing with manual OC just to get an idea of what clocks to voltage ratio it does. I don't want to manual OC due to lack of downclocking/volting? Or can you do it with a P-State like the 1700X? When I tried P-State it wouldn't budge the clock speeds with Minimum processor set to 5%.

I've done PState 0 OC once, it didn't down volt in HWINFO, I can't recall if I checked RM. I will try again as things have moved on in AMD power plan (supposedly). HWINFO is using better method to access CPU so shouldn't send it higher state at idle.

I do feel like the CPU should be doing around 4150MHz in games instead of 4050MHz as with manual OC I can push 4150Mhz at 1.325v so don't understand why it doesn't do that with precision boost.

The SMU has to profile for more reliability than we check a manual OC. So we see what we see....

Anyways I think with BCLK that will be a workaround for now. I don't think it would ever boost as great as yours, but I do think it could be doing a little better. Perhaps future SMU/AGESA revisions might tweak Zen 2 boosting some more. One can hope anyways lol. When I do replace I think I will go for the 3700X or next-gen depending how long it takes me to cave in and buy another lol.

Personally only if you make minimal loss I'd move to another same CPU. If you only game and light productivity, I reckon the R5 3600 is sweet spot.

Again thanks for all the help and putting up with my rambling! I live in a small town and bar my dad there is no one to have tech/hardware discussions lol. So this is my only avenue for said conversations

NP .

[mikev190]

Personally only if you make minimal loss I'd move to another same CPU. If you only game and light productivity, I reckon the R5 3600 is sweet spot.

NP .

I'm just gonna buy a 3600X I think, well in a month or so. There's a few games I want first!

I ran a few CB15 runs and I compared my results to Guru3d's 3600 review and my CPU actually scored better. Not sure how that works

[gupsterg]

I'm just gonna buy a 3600X I think, well in a month or so. There's a few games I want first!

I ran a few CB15 runs and I compared my results to Guru3d's 3600 review and my CPU actually scored better. Not sure how that works

I think it maybe down to AGESA 1.0.0.3 being used in review, link. But again we're gonna see differences with boost, as it's sorta the nature of the beast.

I added 6x more runs to my 0002+ bench album, last 6 is today's runs.

PBO+150MHz with room ambient of ~19-20C can hit ~167x points in CBR15 for me, the 1676 result you see in history is on a "true" AGESA 1.0.0.2 UEFI. The 1610 you see is worst I've ever seen on my rig, AGESA 1.0.0.3AB CPU stock.

If I take the 1610 result, which is worst and was to ref the Kitguru CBR15 results of a 3600X at stock I match it, again AGESA 1.0.0.3 in use for that review.

The CBR15 I did today, actually seems to have hit highest single core result so far, ~205, this bests even manual OC of 4.325GHz on SC, link.

[mikev190]

I think it maybe down to AGESA 1.0.0.3 being used in review, link. But again we're gonna see differences with boost, as it's sorta the nature of the beast.

I added 6x more runs to my 0002+ bench album, last 6 is today's runs.

PBO+150MHz with room ambient of ~19-20C can hit ~167x points in CBR15 for me, the 1676 result you see in history is on a "true" AGESA 1.0.0.2 UEFI. The 1610 you see is worst I've ever seen on my rig, AGESA 1.0.0.3AB CPU stock.

If I take the 1610 result, which is worst and was to ref the Kitguru CBR15 results of a 3600X at stock I match it, again AGESA 1.0.0.3 in use for that review.

The CBR15 I did today, actually seems to have hit highest single core result so far, ~205, this bests even manual OC of 4.325GHz on SC, link.

You have some pretty nice results there! I honestly haven't seen anyone score so decently with a non X 3600.

Most reviews score similar to the screenshot I posted. But what is odd, is that I went back to 0002 with the normal SMU and my scores are the same! So my results are inline with the reviews on the same SMU which makes me feel better. I find it more confusing that the single core score is still the same. The previous SMU was boosting up to 4150MHz. So for now I'm staying with the standard bios as the scores are identical.

I also changed my BCLK to 103 which gains 97Mhz and CB15 score are 1622 multi 198 single so that seems to be a nice bump. I will post some screenshots later!