ROG STRIX X670E-E GAMING WIFI Latest BIOS 0925

[Ryodu]

mine goes up to 1.472v on my 7950X running @ 5725Mhz at 50 degrees. I leave mine on Auto because Manually setting this voltage causes more instability issues due to the fluctuating voltages on the cores running higher clocks

A Vcore of almost 1.5Volts seems to be pretty high to me. Probably more of a problem for my 7950X3D then your 7950X....but not sure

 

[Mussels]

A Vcore of almost 1.5Volts seems to be pretty high to me. Probably more of a problem for my 7950X3D then your 7950X....but not sure

If its only single core at low load, it's not a problem.
Amps is the problem, not voltage.

1.50v for a single core at 20% load is going to use a LOT less amps than 4 8 or 16 cores at the same voltage, and therefore not be an issue.
Amperage creates heat, voltage counters resistance. Amps are the producer of heat, and the danger with such high core count systems.


I've been setting up solar systems for camping recently and that's the first thing they teach you - 24V uses half the amps of 12V, so you can use thinner wiring and have less heat.
This is why it gets so weird with CPU's because technically at the same wattage, a higher voltage would produce less heat and strain on the power wiring and circuitry to the CPU but more heat in the CPU.
This stuff gets complex fast.

 

[mama]

If its only single core at low load, it's not a problem.
Amps is the problem, not voltage.

1.50v for a single core at 20% load is going to use a LOT less amps than 4 8 or 16 cores at the same voltage, and therefore not be an issue.
Amperage creates heat, voltage counters resistance. Amps are the producer of heat, and the danger with such high core count systems.


I've been setting up solar systems for camping recently and that's the first thing they teach you - 24V uses half the amps of 12V, so you can use thinner wiring and have less heat.
This is why it gets so weird with CPU's because technically at the same wattage, a higher voltage would produce less heat and strain on the power wiring and circuitry to the CPU but more heat in the CPU.
This stuff gets complex fast.

I like to think of it as a water pipe analogy where water is amperage and voltage is the pipe. The wider the pipe the more water can flow potentially. Less water in a wider pipe means less chance of a leak. Too much water in a smaller pipe and you have a problem... Now that I've written I think this is a bad analogy.

 

[Mussels]

I like to think of it as a water pipe analogy where water is amperage and voltage is the pipe. The wider the pipe the more water can flow potentially. Less water in a wider pipe means less chance of a leak. Too much water in a smaller pipe and you have a problem... Now that I've written I think this is a bad analogy.

I describe it as a fabric hose, since it works better when you throw voltage into the mix.
I didn't understand electricity at all till I got myself a solar camping setup, learning and working with 12V vs 24V setups finally got it all to click.

There's a reason the engineers use terms like flow rate - because electricity behaves a lot like water.
One key thing that changed my understanding was learning that amps is the sole producer of heat.

Smart phone chargers like the USB-C PD PPS standards can negotiate voltages now, because it lets them send more total wattage over the same cable - a standard USB 2.0 C-C cable can do 3 Amps, so you're stuck at 5v 3A for 15W... unless the devices negotiate for 9V, 12V, etc.
Suddenly you get more power via the same cable with no heat issues - as long as the other end can handle the voltage, and the heat of converting it to a lower voltage internally.

The step-down converter before the battery turns the higher voltage into a lower one at higher amps creating a local heat source they have to keep cool, and as the battery voltage rises in its multiple cells they all gain more pressure to try and escape into nearby cells, resulting in spicy pillows that go boom.
This is where they differ to a CPU, because CPU's dont have those cells storing power physically near each other.

This is why AMD said Amps kills not voltage - because it's only amps that makes the heat. A 5950x can use 4x the wattage of a 5800x safely, they can all run 1.50v safely - but a certain amperage limit makes the CPU overheat, since the heat is concentrated in such a small area (the power pads on the CPU, for example)



Voltage is the pressure, Amps is the flow rate.
Watts could be measured as RPM of a flow meter here - or a combined result of several flow meters, if the powers split to many things.

Low voltage doesnt work well over long distance, as resistance slows the flow rate further.
High amps runs the risk of busting a hole in the hose (cables getting hot, melting, etc)

Sucking more water than the system is designed for makes the pressure drop, and every last thing has to work harder to compensate.
You then have to implement a safety throttle (or many) to cut back the consumption when the flow rate dips to prevent things outright seizing up.


For the CPU Comparison imagine its four hoses going in, and it turns three of them off.
The fourth hose now has four times the pressure and can push one of the four flow meters around faster - but it's still the exact same or even less total water going through.

To achieve the same thing for all paths at the same time however, puts strain on everything upstream - your pump, hoses, joins, connectors, etc
(In the CPU sense the power controller of the CPU, every electrical contact between CPU and socket, the VRMs, power wiring between board and PSU, etc)




Goddamned ADHD, I try and do a simple reply and end up writing Wikipedia articles.

 

[Nam-A-M-D]

Hi All.
I have been testing the system over the last couple of weeks and found over that time i have had to increase the Soc Voltage to 1.250v as the stability did suffer with the occasional blue screen this week. This just shows how temperamental this platform is, but 1.250v seems to be the right voltage for my Motherboard at the moment. You think the system is all good, then all at once a blue screen can cause you to rethink those settings. If they release a new bios soon i will install it to test it out. cheers

 

[Mussels]

Hi All.
I have been testing the system over the last couple of weeks and found over that time i have had to increase the Soc Voltage to 1.250v as the stability did suffer with the occasional blue screen this week. This just shows how temperamental this platform is, but 1.250v seems to be the right voltage for my Motherboard at the moment. You think the system is all good, then all at once a blue screen can cause you to rethink those settings. If they release a new bios soon i will install it to test it out. cheers

SoC voltage has always been required to keep the system stable at *low* load, as it's the transition states between idle and load that tax the motherboards power delivery the hardest

It has to calm down to not roast the CPU at idle, then rapidly ramp up in under a millisecond to full power and thats when the voltages droop - and depending on what you're doing, it can transition back and forth a *LOT* very very fast.
Ryzen 3000 introduced the 1ms polling rates, so it's easy to see why these SoC VRM's have trouble preventing voltage droops when the spec is for 1.1v? (I forgot AM5's stock) and people need it stable at 1.20-1.25v for 6000MT/s, with such rapid changes to power delivery. Despite being overclocking boards with "upto" X DDR5 speeds, they used the dirty hack of "over volt to compensate" instead of better regulation of the SoC voltages

The short-term solution is to not run your ram as fast and use lower speeds with tighter timings with a lower SoC voltage, the long term solution is for motherboards to use better VRM's for the SoC so they can prevent droop properly instead of just blasting higher voltages.

It's like the dream of DDR4 4000 on AM4, eventually everyone understood that it wasn't worth it and settled for 3600-3800.
AM5 still has this obsession with DDR5 6000 and above, and they just cant do it easily.

 

[Nam-A-M-D]

SoC voltage has always been required to keep the system stable at *low* load, as it's the transition states between idle and load that tax the motherboards power delivery the hardest

It has to calm down to not roast the CPU at idle, then rapidly ramp up in under a millisecond to full power and thats when the voltages droop - and depending on what you're doing, it can transition back and forth a *LOT* very very fast.
Ryzen 3000 introduced the 1ms polling rates, so it's easy to see why these SoC VRM's have trouble preventing voltage droops when the spec is for 1.1v? (I forgot AM5's stock) and people need it stable at 1.20-1.25v for 6000MT/s, with such rapid changes to power delivery. Despite being overclocking boards with "upto" X DDR5 speeds, they used the dirty hack of "over volt to compensate" instead of better regulation of the SoC voltages

The short-term solution is to not run your ram as fast and use lower speeds with tighter timings with a lower SoC voltage, the long term solution is for motherboards to use better VRM's for the SoC so they can prevent droop properly instead of just blasting higher voltages.

It's like the dream of DDR4 4000 on AM4, eventually everyone understood that it wasn't worth it and settled for 3600-3800.
AM5 still has this obsession with DDR5 6000 and above, and they just cant do it easily.

I agree that it is the drooping power at low load what caused the Blue screen and i have compensated this issue and lowered the speed until we get a baseline that is stable. Cheers

 

[davfel01]

please musselman stop making things up.
just quote one source of anything you say.
just one

 

[Mussels]

please musselman stop making things up.
just quote one source of anything you say.
just one

Try clicking the blue text with all the hyperlinks involved. I always post links and quotes, it's not my fault if you don't check them.

Then if you feel like dragging things off topic more, I'm happy to give out infractions.

 

[eidairaman1]

please musselman stop making things up.
just quote one source of anything you say.
just one

How about you?

 

[davfel01]

which would you like a reference too?

that amps aren't heat? try ohm's law

 

[Nam-A-M-D]

JUST RELEASED

ROG STRIX X670E-E GAMING WIFI BIOS 1516 BETA

Version 1516 Beta Version
9.23 MB 2023/07/24
"1. Update AGESA version to Combo AM5 PI 1.0.0.7b
2. Improve memory compatibility

Some of the changes.

AGESA 1.0.0.7B BIOS improves DDR5 memory support. The previous reports indicate that the previous limit of DDR5-6000 has been increased significantly to DDR5-7000 and even DDR5-8000.

AMD has massively improved the DDR5 memory support on the AM5 platform with a new microcode update known as AGESA 1.0.0.7b. The update allows some boards to push DDR5 frequencies as high as 8000MHz without issue and boosts the optimal 1:1 (UCLK:MEMCLK) ratio to 6400MHz. Gigabyte and ASRock have already published stability runs with the new BIOS update, showing off 8000MHz and 7200MHz working perfectly on the Aorus X670E Master and X670E Taichi. I will be interested to try out the 8000Mhz Memory to see how it compares to what i have now in speed performance on games and Rendering Production.


Only a few AM5 motherboards have the new microcode update for now. Still, we suspect the update will become more readily available over the next couple of weeks, depending on the launch schedule of each motherboard manufacturer.

 

[ir_cow]

I'm waiting for the Hero to get this B microcode update. Didn't see the GB Master one available yet publicly. Only ASRock and ASUS for one MB so far. I say give it a few weeks and they all have it.

GEAR 2 has a fair bit of performance drop. Really not worth being excited over unless you're running over DDR5 8000. Even than DDR5-6000 CL30 will beat it I'm Gear 1

 

[Mussels]

I'm waiting for the Hero to get this B microcode update. Didn't see the GB Master one available yet publicly. Only ASRock and ASUS for one MB so far. I say give it a few weeks and they all have it.

GEAR 2 has a fair bit of performance drop. Really not worth being excited over unless you're running over DDR5 8000. Even than DDR5-6000 CL30 will beat it I'm Gear 1

Please. PLEASE.

I need to see you post DDR5

 

[ir_cow]

@Mussels LOL you will probably long before I do.

 

[Nam-A-M-D]

I have to agree that unless you are planning on using DDR 8000 Memory it isn't worth the messing around unless there are any under the hood tweaks they have done without reporting. I will check the update out for any performance increases and let everyone know the outcome. Hopefully in the coming weeks i will be getting some DDR 8000 to test out fully. Cheers

​

 

[Nam-A-M-D]

Latest Bios Update
ROG STRIX X670E-E GAMING WIFI BIOS 1602

Version 1602
9.24 MB 2023/08/18
"1. Update AGESA version to Combo AM5 PI 1.0.0.7c
2. Improve system performance

Before running the USB BIOS Flashback tool, please rename the BIOS file (SX670EE.CAP) using BIOSRenamer."

ROG STRIX X670E-E GAMING WIFI | ROG Strix | Gaming motherboards｜ROG - Republic of Gamers｜ROG Global

ASUS ROG Strix X670E-E Gaming WiFi features 18+2 power stages, DDR5 support, four M.2 slots with heatsinks, PCIe® 5.0, USB 3.2 Gen 2x2 front-panel connector, and onboard WiFi 6E. It also features AI Cooling II, AI Networking, Two-Way AI Noise Cancelation. ROG Strix X670E-E Gaming WiFi caters to...

rog.asus.com

The last revision bios in July seems to have caused stability issues from the memory, which causes random crashes from time to time whilst using different apps. Also i have noticed when running on a cold boot the system will hang after loading onto the desktop then restarting again with no problems. Measuring the time on the drivers loading this could be a Wifi 6e adapter driver problem. The Wifi 6E would randomly drop out and disappear from the operating system and the only way to get this to work again is to reinstall the drivers. I have been investigating all these issues since July and i hope this revision resolves the problems. I will let you all know how it goes. cheers

 

[Mussels]

The last revision bios in July seems to have caused stability issues from the memory

They lowered SoC voltage from prior ones, because the prior settings was exploding CPUs. You need to correctly set your SoC voltage manually.

 

[Nam-A-M-D]

Hi. I manually set the voltage to 1.3v again having the startup issues, but it did still cause the restarts. This morning after updating to the new bios it seems to be ok again. I will see how it goes. Cheers

 

[ir_cow]

Hi. I manually set the voltage to 1.3v again having the startup issues

1.3 SoC is the MAX now. Above and you risk a meltdown situation. really shouldn't need this much for normal stuff.

 

[Mussels]

Hi. I manually set the voltage to 1.3v again having the startup issues, but it did still cause the restarts. This morning after updating to the new bios it seems to be ok again. I will see how it goes. Cheers

testing at 1.30 was a good start, but make sure you test lower too. 1.25, 1.20, 1.15.

Higher isn't always more stable, but it is valid to start with a higher range and work down.

 

[Nam-A-M-D]

I set the voltage 1.275v in the end like I did on previous bios revisions previously in this thread. Setting it to low caused issues, but way back in this thread i can remember lowering it even further but it depends on so many factors and settings. cheers

 

[ir_cow]

I set the voltage 1.275v in the end like I did on previous bios revisions previously in this thread. Setting it to low caused issues, but way back in this thread i can remember lowering it even further but it depends on so many factors and settings. cheers

Don't forget about VDDIO-MEM voltage. Might just need to bump that and SA can be lower.

 

[Steevo]

I'm waiting for the Hero to get this B microcode update. Didn't see the GB Master one available yet publicly. Only ASRock and ASUS for one MB so far. I say give it a few weeks and they all have it.

GEAR 2 has a fair bit of performance drop. Really not worth being excited over unless you're running over DDR5 8000. Even than DDR5-6000 CL30 will beat it I'm Gear 1

I.have 64GB of ram running at tight timings on gear 2 and when I try gear 1 I only gain 2ns better latency and a but more bandwidth, and BSOD at stock.settings.

 

[ir_cow]

I.have 64GB of ram running at tight timings on gear 2 and when I try gear 1 I only gain 2ns better latency and a but more bandwidth, and BSOD at stock.settings.

Tight timings can make up for some performance loss in Gear 2, but Gear 1 at a lower frequency and equally or better timings will certainly beat it in synthetic benchmarks.

Real world, well... If your GPU bound in games, besides potentially higher 1% lows, it won't be much different.