ASUS ROG Crosshair VIII Hero (WiFi)

[TheLostSwede]

I am well aware of that but again I must come back (my opinion) why install x16 slots on the board if only 1 can run at that speed. Especially at the price point for some of these boards, it would seem ridiculous to me that you have slots that you can't be fully used, that is why I never got into B450 boards.

Because SLI/CrossFire.
Also, because the length of the physical slot has nothing to do with the interface bandwidth available.
Also, PCIe 4.0 is much faster than 3.0, so once there are more 4.0 devices, each device will technically step down in terms of the physical slot needed, so current x4 PCIe 3.0 cards will get enough bandwidth from a x1 PCIe 4.0 slot.
Also, the third x16 slot is normally x4 in terms of bandwidth (not sure what Asus did here, it's always x4 on my board) and that was something that started on Intel boards if I'm not wrong here...

 

[kapone32]

Because SLI/CrossFire.
Also, because the length of the physical slot has nothing to do with the interface bandwidth available.
Also, PCIe 4.0 is much faster than 3.0, so once there are more 4.0 devices, each device will technically step down in terms of the physical slot needed, so current x4 PCIe 3.0 cards will get enough bandwidth from a x1 PCIe 4.0 slot.
Also, the third x16 slot is normally x4 in terms of bandwidth (not sure what Asus did here, it's always x4 on my board) and that was something that started on Intel boards if I'm not wrong here...
[/QUOTE]

While I do admire and use SLI/crossfire so I get what you are saying. Instead of making them seem like full 16 slots why did they not make the 3rd slot a x4 as it is wired. Yes 4.0 is in theory 2x faster than 3.0. On the Ryzen MBs it has always been that the 3rd slot is wired at x4, on some PCI_E 2.0x4 as well. It would not be for you and I (frankly most people on the forum) In a world of capture cards and expansion cards it would be easy for a noob who wants to stream to Twitch or Youtube to get a capture card and an expansion card thinking they will work. I personally do not see a difference between the X570 and upcoming B550 boards in terms of usability, for the 16 slots. There is nothing gained using a GPU that is 4.0 right now in terms of performance, In theory vendors could release a board wired at 3.0 with 3 full PCI_E 3.0x16 slots (Biostar maybe) In fact I would rather see that than the current crop of X570 boards. There is nothing gained using a GPU that is 4.0 right now in terms of performance and unfortunately you would not be able to use a x4 interface in a x1 slot even though it would be speed compliant. There is also the fact that the new GPU thought process (make the card as wide and long as possible) may exclude you from being able to use the 2nd slot anyway.

 

[Oberon]

So the X570 boards do indeed come with PCI_E lane splitting?

Yes; how else would you get the x8/x8 and x8/x4/x4 configurations? You can see where it's configured in BIOS in the review, e.g.:

I am well aware of that but again I must come back (my opinion) why install x16 slots on the board if only 1 can run at that speed. Especially at the price point for some of these boards, it would seem ridiculous to me that you have slots that you can't be fully used, that is why I never got into B450 boards.

It's probably a combinations of a few things. It looks more "premium" vs having a x8 and x4 slot on the board, and it probably serves to reduce consumer confusion (or not, as it seems) vs having open-ended x8 and x4 slots on the board that still accept PCIe cards with longer connectors.

 

[Tomgang]

Oh a review of the board i am getting for my next CPU upgrade. How exiting and cut not care less about RGB light as i turn it of any way. Hopes its ready for the Ryzen 9 3950X.

 

[Nkd]

what's the deal with Far Cry 5 though? surprised to see such a big difference between motherboards?

View attachment 133312

Its because x570 are all tested with ddr4 2400 in reviews at TPU. Thats the reason.

 

[bug]

Seems a bit of a cop-out, basicly saying because something takes time to re-do or update, you should NEVER re-do or update anything, because it takes time.

Does he not get paid to do the reviews?, if not? - alright, fine.

It's not copping-out. The reference testing platform gets upgraded from time to time. Just not when you want.

 

[Axaion]

It's not copping-out. The reference testing platform gets upgraded from time to time. Just not when you want.

If its out of date, its out of date. its obviously out of date - youre not fooling anyone.

 

[claster17]

About that chipset fan. Mine is clearly audible in a silent system but thankfully it never turns on.

It's probably a combinations of a few things. It looks more "premium" vs having a x8 and x4 slot on the board, and it probably serves to reduce consumer confusion (or not, as it seems) vs having open-ended x8 and x4 slots on the board that still accept PCIe cards with longer connectors.

It's cheaper to order multiple x16 slots instead of ordering separate x16, x8 and x4.
Those shorter slots also have less structural support in case you want to stick a heavy graphics card in there.

 

[Black Haru]

I'm pretty sure the fans are inaudible out of the box, otherwise reviewers will slam these hard. The problem, however, is how it behaves a year or two down the line. Because it's a mechanical moving part and prone to collecting some dust.

I would be curious to here some long term reports from users. In testing, even on an open bench the only time I hear the fans is on first boot up (most boards will spin the fan to 100% for a few seconds).

what's the deal with Far Cry 5 though? surprised to see such a big difference between motherboards?

View attachment 133312

The difference between X570 and Z390 has more to do with the CPU. A Ryzen 5 3600X is not on par with an i9 9900k in games (I'll have the platforms color separated for easier reading in the next review)
The ~4 fps difference in X570 boards is basically margin of error.

Its because x570 are all tested with ddr4 2400 in reviews at TPU. Thats the reason.

The Intel reviews use literally the exact same memory kit....

If its out of date, its out of date. its obviously out of date - youre not fooling anyone.

Testing updates have to be done Before new hardware launches. Launch windows are incredibly tight as it is just with testing the new boards. So new testing standards will always be a generation behind unless absolutely vital. While using 3200 MHz memory would change benchmark numbers, it is a stretch to assume the numbers would change much relative to one another (beyond margin of error).

If I am mistaken, please help me understand why this is so important, I take pride in my work; just understand that what you are asking for is dozens of hours of completely unpaid labor.

 

[TheMadDutchDude]

I’ve been using it since launch day and it’s just as quiet as it first was. Actually, it’s quieter now that they’ve reduced the fan curve at idle. I wish you could control it in the UEFI, though.

 

[Rebe1]

I bought two months ago a half-year old C7h Wifi for decent price. Was super excited as this was my very first high-end mobo. And what I see when my 3700x finally booted? Async BCLK generator not available for Picasso, max boost 4350 MHz, "cold boot" issue at startup. It took Asus nearly 3 months to fix boost clock, PBO still does not work (have AiO), "cold boot" still present.

FU, @Asus. Never more.

 

[EarthDog]

The ~4 fps difference in X570 boards is basically margin of error.

Is it?

That is a fair amount. If you are talking 100 fps, 4 is 4% difference. Most consider run variance/moe 1%...maybe 2%.

And if that is around 60 fps its 7%....

It's not copping-out. The reference testing platform gets upgraded from time to time. Just not when you want.

The point is this was started, essentially, in error. Even though the difference is a couple percent, it's still a difference and still running the platform WAY under platform spec.

We went over this already in a previous review and left it scratching our heads. It could have been done back then and not rerun as many...but the process soldiered on..

 

[Axaion]

I would be curious to here some long term reports from users. In testing, even on an open bench the only time I hear the fans is on first boot up (most boards will spin the fan to 100% for a few seconds).



The difference between X570 and Z390 has more to do with the CPU. A Ryzen 5 3600X is not on par with an i9 9900k in games (I'll have the platforms color separated for easier reading in the next review)
The ~4 fps difference in X570 boards is basically margin of error.



The Intel reviews use literally the exact same memory kit....



Testing updates have to be done Before new hardware launches. Launch windows are incredibly tight as it is just with testing the new boards. So new testing standards will always be a generation behind unless absolutely vital. While using 3200 MHz memory would change benchmark numbers, it is a stretch to assume the numbers would change much relative to one another (beyond margin of error).

If I am mistaken, please help me understand why this is so important, I take pride in my work; just understand that what you are asking for is dozens of hours of completely unpaid labor.

Sure, it takes time to update testbench and setups, but it -does- have to be done at some point.

As im sure youve read or seen HUB/GN reviews, it impacts 1%, and 0.1% lows quite a bit, at least for ryzen systems, probably for intel too?, dont know - didnt see the tests for those.

You specify its "2x 8 GB DDR4 3866 MHz" ram sticks, so that wouldent be the limiting factor, something like 3200mhz cl14 should be a cakewalk for them at the very least.
It does scale with other speeds/timings however, referring back to hub/gn
Specially seeing as IF speeds are (usually) tied to memory clock's.


Its not only motherboard reviews that have this problem, case reviews dont to temp testing either, so theyre pretty much more of very nice unboxings than reviews.


But yeah, if youre unpaid for the work, why should you?, i wouldnt. - but you cant have a comment section and expect people to not comment, or pitch in what could be improved, or what is lacking.

Just commenting as someone who used to have TPU as my go-to place for numbers reviews, now i use it more as a news site really.

 

[IceShroom]

I still don't understand why you keep using ddr4 2400 mhz. I mean are you kidding me? 90ns memory? lol. So you think someone is going to grab a board for 400 and throw in ddr4 2400 mhz? Atleast test the thing with minimum official support. I am running 3733mhz and I have 64-65ns in AIDA64 and way better results then you in this review. I am not sure if TPU is doing this to put x570 in bad shape on purpose, lol.

Especially when you have DDR4 3200 1.2V for sale.

Crucial 8GB DDR4 3200 (PC4 25600) Desktop Memory Model CT8G4DFS832A - Newegg.com

Buy Crucial 8GB DDR4 3200 (PC4 25600) Desktop Memory Model CT8G4DFS832A with fast shipping and top-rated customer service. Once you know, you Newegg!

www.newegg.com

I'm sure Nate will be happy to do it once you volunteer to re-run all the previous tests for him.

@W1zzard has no problem testing every gpu. Besides there are only few cpu to test.

 

[bug]

I would be curious to here some long term reports from users.

Good luck finding a long term report with the product out for only 3 months.

In testing, even on an open bench the only time I hear the fans is on first boot up (most boards will spin the fan to 100% for a few seconds).

I think all fans do that on startup. Annoying, but better than frying something before the UEFI can figure out who's who

 

[crotach]

What are all these other systems you compare it to? Z390 is much faster in gaming, but is it i9 9900k or what?

 

[Jism]

In the OC test; you guys always seem to put 1.4V as a default voltage into the CPU's. Is'nt the Ryzen 2x00 and Ryzen 3x00 series prono to faster degradation with such voltages? You guys should know that the max 24/7 voltage for those 3x00 series is around 1.37v or so.

 

[Oberon]

In the OC test; you guys always seem to put 1.4V as a default voltage into the CPU's. Is'nt the Ryzen 2x00 and Ryzen 3x00 series prono to faster degradation with such voltages? You guys should know that the max 24/7 voltage for those 3x00 series is around 1.37v or so.

The problem with that is that it simply isn't true. For manual overclocking 1.4 - 1.45 V is fine for daily use as long as temperatures are kept under control. This is as stated by officials from AMD as well as motherboard makers, as opposed to all of the process engineers from Armchair University.

 

[EarthDog]

The problem with that is that it simply isn't true. For manual overclocking 1.4 - 1.45 V is fine for daily use as long as temperatures are kept under control. This is as stated by officials from AMD as well as motherboard makers, as opposed to all of the process engineers from Armchair University.

links please..

 

[Oberon]

Gigabyte's statement is contained in the linked 3950X overclocking guide here. As for AMD's stuff, you'll have to go dig through their community updates/reddit posts about voltages.

 

[Jism]

The problem with that is that it simply isn't true. For manual overclocking 1.4 - 1.45 V is fine for daily use as long as temperatures are kept under control. This is as stated by officials from AMD as well as motherboard makers, as opposed to all of the process engineers from Armchair University.

https://www.reddit.com/r/Amd/comments/8cppib
https://www.reddit.com/r/Amd/comments/amvfi4
https://www.reddit.com/r/overclocking/comments/bbxot0
This is just the top 3 selection on Google. Yes Ryzen suffers from fast degradation once voltages over 1.4V applied. I find it so stupid that many reviewers dumbly put the voltage manually on 1.4V ~ 1.45V and start crunching things like CB or Linpack for that matter. These CPU's should not go over 1.34v (2x series) and 1.37v (3x series) as tested by quite a few people.

The problem is with "good" cooling is that 7nm brings something to the table and that is a enormous heat density. It's difficult to keep that under 60 degrees. Even my 360mm rad with 6 fans has troubles cooling a stock Ryzen 2700x. It has to have a slight undervolt to stick at 4.2Ghz all core boost.

 

[Oberon]

As if any of these people are qualified to assess whether they have experienced electromigration or any other form of degradation of their chips...

 

[Jism]

linked

Oh well i suggest you cram up to 1.45v into your 3x00 series, let me know how it goes after 24 hours prime (Y)

 

[EarthDog]

As for AMD's stuff, you'll have to go dig through their community updates/reddit posts about voltages.

Nope. That is for you to find since you brought it up. Also, the Giga guide doesn't explicitly mention a maximum voltage (did I miss it?), just what they suggest for a 4 GHz overclock in the Vcore section.

Hit us back up when you have the AMD link please.

The reality of things is, for the most part, over 1.35V can't be cooled by a 3x120mm AIO (using a stress test and keeping it at or under 90C). Perhaps this has changed with the better binned 3950x as we saw from the Giga guide and THAT sample, but anything else out there, 1.35V is about it for ambient cooled thermals in the first place...

 

[Oberon]

https://www.reddit.com/r/Amd/comments/cbls9g
There you go. Specifies up to 1.5 V for boost behavior. That does NOT state 1.5 V for 24/7 usage, but A) you aren't running 1.5 V for a manual overclock because it doesn't gain you anything, and B) if one core can handle it they all can. The cooling bit is a red herring. If you can't cool it, get better cooling. 360mm AIOs are still in the realm of budget cooling if you're actually chasing max overclocks.

As for the Gigabyte thing, you might want to read it again:

We recommend you to stay under 95° C on your CPU along with a CPU Vcore below 1.45 volts if possible.

Note: that's not a recommendation for a particular speed; that is a general recommendation for adjusting voltages given further down in the guide.

*EDIT* Good lord, those automatic media links are annoying...