ASUS Z87 Motherboards - Overview, Guides and Official Support

[RPGWiZaRD]

HERO has amazing audio, better than Extreme6 in my testing. Working on my review of the Extreme6 right now, actually. I have some performance issues with EX6 I am trying to figure out, OC isn't holding what I set and I see multi drop about 300 MHz like it's running default Intel Turbo with raised multis. Could be a cooling thing for the board since I pulled VRM coolers for VRM pics, but I replaced and contact seems good. Still trying to figure it out, honestly, but today is the last day I spend on it. See anything like that with yours?

My HERO review is here, so you can check RMAA results for HERO:

http://www.techpowerup.com/reviews/A...XIMUS_VI_HERO/

RMAA results for EX6 has high SNR, but also, more noise.

Thanks, did you happen to notice any differences in how much EMI / interfearence the boards pick up? The ASRock Extreme6 to me has unacceptably much EMI noise (HDD activity, mouse interfearence, high system load etc). It's one of the reasons I wanna swap boards in fact, I'm using a headphone amp so ofc it will be worse then but I love the sound of the amp so I'd just prefer to find an onboard solution (I also happen to like how the soundstage sounds like on Realtek when set to 5.1 speakers when using headphones) that just has low enough static interfearence from the rest of the components, at least as low as my ALC889A chip I had before... it wasn't actually low but yea much lower than with the ASRock board at least... Getting popping sound often when minimizing & maximizing a window and the HDD activity can be heard a bit with a buzzing/popping sound (depends on the vol set on the amp, at full volume and lower Realtek output vol which was borderline acceptable on the old ALC889A chip but very unacceptable on the Extreme6, so have to lower the amp volume a lot with this Extreme6 but it is still very much there and then it also sounds a bit worse than if using a bit higher vol setting on the amp and lower vol on the source so basically I'd want to be able to go high on the amp vol without getting that noise).

And yes I've tried 3rd party soundcards including ASUS D2, Titanium HD and SoundBlaster ZxR but I just prefer the way soundstage is portrayed with onboard (when using 5.1 speakers with my headphones), the ZxR while having excellent quality especially in the midrange sounds, has very closed-in sound compared to onboard no matter how the speaker config is setup.

EDIT: I haven't specifically payed much attention to the performance yet but it feels a bit sluggish at times maybe. Running at a modest 4.4GHz only but it does so with ease though and it seemed much more difficult to get 4.5GHz stable. But with the ASUS board I will very likely give another go at trying to push to 4.5-4.6GHz as the temps are very low still.

[cadaveca]

With how I test, noise and interference as well as interchannel leakage is about 100x better(literally) on the HERO vs EX6, and HERO is better at audio reproduction, too. The EX6 is a few dB louder.

I haven't used the HERO too much, honestly, just while doing the review, so a couple of weeks. I definitely notice the difference between the two, but I'm kinda picky about sound..not in an audiophile way, but as a musician. So what's good to me, might not exactly be what you are looking for...but I can say that the noise difference(one of the main things I noticed) is HUGE.

[RPGWiZaRD]

With how I test, noise and interference as well as interchannel leakage is about 100x better(literally) on the HERO vs EX6, and HERO is better at audio reproduction, too. The EX6 is a few dB louder.

I haven't used the HERO too much, honestly, just while doing the review, so a couple of weeks. I definitely notice the difference between the two, but I'm kinda picky about sound..not in an audiophile way, but as a musician. So what's good to me, might not exactly be what you are looking for...but I can say that the noise difference(one of the main things I noticed) is HUGE.

OK, that's good =) Didn't expect much about ppl able to comment on audio here as A) this is XtremeSystems and B) Realtek onboard audio is usually sneezed at xD Would be cool to know for certain if GENE and HERO have pretty similar results or if one or the other is better in this regard. At least ASUS says GENE has 115dB SNR... equally high as Extreme6 so they should be roughly equally loud before distortion kicks in or the noise floor is lower, it doesn't say for HERO, usually doesn't say means it's usually not as good (doesn't have to be like that but often is...)

And no I don't mean higher SNR rating automaticly would mean less noise or better quality but I'm interested in if there's ANY difference (opamp or caps or whatever) between HERO and GENE which differing specs could probably reveal. Then I could dig up more info on those components to know what should at least theoretically have less noise looking at the paper specs...

[cadaveca]

OK, that's good =) Didn't expect much about ppl able to comment on audio here as A) this is XtremeSystems and B) Realtek onboard audio is usually sneezed at xD Would be cool to know for certain if GENE and HERO have pretty similar results or if one or the other is better in this regard. At least ASUS says GENE has 115dB SNR... equally high as Extreme6, it doesn't say for HERO, usually doesn't say means it's usually not as good (doesn't have to be like that but often is...)

And no I don't mean higher SNR rating would mean less noise or better quality but I'm interested in if there's ANY difference (opamp or caps or whatever) between HERO and GENE which differing specs could probably reveal.

I never reach the SNR that most OEMs report. I do have ASUS's guide on how to replicate their results, but prefer my own method, which seems pretty good at picking out actual noise from other bits on the board(although I do test in basically the same way they tell reviewers to).

I do have the GENE as well, but haven't got to it yet, and might not for a couple of weeks yet. I'm not the sort to comment on stuff unless I have the hardware in my hands and have tried it, so I cannot offer you any info about the GENE vs the HERO. I haven't even taken the GENE out of the box yet, and wasn't planning to until August. Oh crap..it's almost August now, isn't it? LOL.

[RPGWiZaRD]

I do have the GENE as well, but haven't got to it yet, and might not for a couple of weeks yet. I'm not the sort to comment on stuff unless I have the hardware in my hands and have tried it, so I cannot offer you any info about the GENE vs the HERO. I haven't even taken the GENE out of the box yet, and wasn't planning to until August. Oh crap..it's almost August now, isn't it? LOL.

Damn, you wouldn't be able to speed up the review of GENE would you!? Don't know if I could wait for weeks, I'd pretty much wanna get a new mobo ASAP haha. I'd be pretty devastated if I afterwards got to learn GENE actually produces a bit better results than HERO and both boards costs the same etc.... :/

[Eldonko]

Hey Dave, what do the RMAA results look like for Extreme 6? I would like to see side by side vs Hero, do you have that?

[cadaveca]

Hey Dave, what do the RMAA results look like for Extreme 6? I would like to see side by side vs Hero, do you have that?

Of course. You can check the HERO review for the HERO results, here's the EX6, although the review isn't live yet. Will be soon.:

[Eldonko]

Wow, pretty big difference. Thanks.

[cadaveca]

NP. It's not that that ASRock board has bad audio, for me, it's about average for decent onboard, rather than just normal on-board.



But the HERO audio is just truly exceptional.

[radlord]

I think i might have a problem with UEFI settings, namely the HDD boot orders (it wont stick to my SSD and gets stuck in a boot/reset loop for ever until you catch it and go into the UEFI).

Also, somewhat less importantly - but still annoyingly as I have Windows 8 set to require a PIN at logon - the numlock doesn't default to 'on' even though this is set in BIOS 1205.

[cstkl1]

@Raja @LardArse

Is there a way to fix the tWTR. Set 9 and get 10. So far Arock Timing Configurator shows all the settings including tWR and tWTR.

btw is there any formula for tCPDEN/tRDPDEN and whats cmd_strech?? Anyway to change both??

[RPGWiZaRD]

Of course. You can check the HERO review for the HERO results, here's the EX6, although the review isn't live yet. Will be soon.:

Wow that is quite depressing THD noise figure on the ASRock Extreme6, I dug up some measurements on ALC889A, which my previous board had and it had like -76~78 dB(A) depending on settings used. Guess this explains the EMI too with the ASRock board. They have very good Dynamic Range (probably top-of-the-line onboard-wise without going to special "Sound Core3D" chips / Creative onboard solutions which seem to give an additional or so 3-5dB dynamic range. Seems like the ASRock implementation is rather shoddy, they only cared about getting some nice output figures but forgot about the noise....

[Raja@ASUS]

@Raja @LardArse

Is there a way to fix the tWTR. Set 9 and get 10. So far Arock Timing Configurator shows all the settings including tWR and tWTR.

btw is there any formula for tCPDEN/tRDPDEN and whats cmd_strech?? Anyway to change both??

think Shammy alread said these are offset so no reported in UEFI - that's how it actually works. OS reading could be bugged or whatever as well.

As for tCPDEN etc - these are power saving features, time from a transaction to power down. If you disable all power saving features of DRAM I wouldn't bother with this kind of "tweaking" myself. I am being sarcastic with that comment, but I think you'll get what I mean :P

The salient tweak features of this chipset are already pointed out in the guides. If you want to move outside that - don't expect much if any return for your efforts.

[Raja@ASUS]

I never reach the SNR that most OEMs report.

http://www.ap.com/


A dedicated audio analyzer from those guys is what one needs to measure this stuff properly - unfortunately it falls beyond the scope of most can use etc. Thats what we use at HQ to measure our stuff for the SNR figures in the marketing materials.

RMAA loopback testing is limited by the input SNR/dynamic range as well - so there will always be offset between the actual SNR/DR and what one can get with RMAA. The guides on RMAA are written so one can set the software up and at least get to the best possible measurement - hardware/software limitations of this form of testing withstanding.

-Raja

[Praz]

[@Praz][@Dumo][@owikh]

At what ram speed u have to up the tRDRD from perfect 4.

I think Raja has already stated somewhere that most memory is going to top out around 2400MHz with tRDRD at 4. Testing the tRDRD value against memory speed will need to be done for max performance. If a system can only reach 2300MHz memory speed because of tRDRD being at 4 and setting it to 5 allows for 2350MHz this tradeoff will not be worthwhile. Performance will surely be better at 2300MHz memory speed and tRDRD at 4. However if changing only tRDRD to 5 allows the memory to be run at say 2500MHz this increase in memory speed will provide more performance than the 1 clock hit of tRDRD .

[cstkl1]

think Shammy alread said these are offset so no reported in UEFI - that's how it actually works. OS reading could be bugged or whatever as well.

As for tCPDEN etc - these are power saving features, time from a transaction to power down. If you disable all power saving features of DRAM I wouldn't bother with this kind of "tweaking" myself. I am being sarcastic with that comment, but I think you'll get what I mean :P

The salient tweak features of this chipset are already pointed out in the guides. If you want to move outside that - don't expect much if any return for your efforts.

Ok got it but found how to change the twtr already flukishly.both of u helped alot so far... So the dream is to have a 40k read/write/copy so just trying to get a understanding on the timings to solve the 2600 1t trdrd4 issue. 2t np fully stable.

I think Raja has already stated somewhere that most memory is going to top out around 2400MHz with tRDRD at 4. Testing the tRDRD value against memory speed will need to be done for max performance. If a system can only reach 2300MHz memory speed because of tRDRD being at 4 and setting it to 5 allows for 2350MHz this tradeoff will not be worthwhile. Performance will surely be better at 2300MHz memory speed and tRDRD at 4. However if changing only tRDRD to 5 allows the memory to be run at say 2500MHz this increase in memory speed will provide more performance than the 1 clock hit of tRDRD .

I got it up to 2600 c10 but in 2t fully stable. 1t is being at odds with me but some progress on it. hence needed some lines on where it taps out...since u guys have tested more kits and samsung 2600 n 2666 in particular

[Raja@ASUS]

Ok got it but found how to change the twtr already flukishly.

Change values till you get what you want. Min internal spacing for this will be 4 clocks - not many DIMMs will do that at high freq.

[cstkl1]

think Shammy alread said these are offset so no reported in UEFI - that's how it actually works. OS reading could be bugged or whatever as well.

As for tCPDEN etc - these are power saving features, time from a transaction to power down. If you disable all power saving features of DRAM I wouldn't bother with this kind of "tweaking" myself. I am being sarcastic with that comment, but I think you'll get what I mean :P

The salient tweak features of this chipset are already pointed out in the guides. If you want to move outside that - don't expect much if any return for your efforts.

Change values till you get what you want. Min internal spacing for this will be 4 clocks - not many DIMMs will do that at high freq.

nah i always like my trtp and twtr to be the same. Not aiming for minimum. Cause at default theres a issue with memtest, on cold boot it always shows error on test 1. But rerun or restart it disappears. Changing this solved it. Just seriously want 2600 1t.

Btw thanks for making this board. Its awesome.

My biggest headache with haswell is digital i/o n twrwr_dr/dd. this is whats stopping me atm. Also need this to be stable to have a shot at 2666 . Personally gut feeling is telling me this is where trdrd 4 taps out.

If any board can do it .. It will be m6e at 4x4gb. The exact settings to make 2400mhz 1t work is what did the trick for 2600 2t. So hoping 2600 1t will do for 2666 2t

[Raja@ASUS]

nah i always like my trtp and twtr to be the same.

You are confining yourself for no reason there. Unless you keep both at 4 clocks (which is simply their minimum time based upon burst lengths), there is no correlation to keep these the same for peace of mind.

PRECHARGE closes a page, tRTP sets the delay between the last read transaction and tRP. tWTR is simply the delay between a write transaction and a read. The two settings are not mutually related in any way that would require you to confine yourself to the rule you state.

Anyway, the rest of your dimlemma is simply a matter of taking your time and working out what's right for you - don't worry about what other people can or can't run. IO can depend on many things so one is best advised to tune themselves gradually if so inclined. One persons stable is another persons unstable too - I have heard of people running tRDRD at 4 over DDR3-2800 on Ln2, does not mean we can all do it. For 24/7 systems most people will find that stress tests will start failing past DDR3-2400 if the bus is loaded - both the DIMMs and memory controller will need more overvoltage than many are willing to run on their systems to go any higher even if it were possible. Plus, the performance change in everyday systems is too small for many people to worry about. The only place this stuff starts to matter a lot is Super Pi 32M.

[cstkl1]

You are confining yourself for no reason there. Unless you keep both at 4 clocks (which is simply their minimum time based upon burst lengths), there is no correlation to keep these the same for peace of mind.

PRECHARGE closes a page, tRTP sets the delay between the last read transaction and tRP. tWTR is simply the delay between a write transaction and a read. The two settings are not mutually related in any way that would require you to confine yourself to the rule you state.

Anyway, the rest of your dimlemma is simply a matter of taking your time and working out what's right for you - don't worry about what other people can or can't run. IO can depend on many things so one is best advised to tune themselves gradually if so inclined. One persons stable is another persons unstable too - I have heard of people running tRDRD at 4 over DDR3-2800 on Ln2, does not mean we can all do it. For 24/7 systems most people will find that stress tests will start failing past DDR3-2400 if the bus is loaded - both the DIMMs and memory controller will need more overvoltage than many are willing to run on their systems to go any higher even if it were possible. Plus, the performance change in everyday systems is too small for many people to worry about. The only place this stuff starts to matter a lot is Super Pi 32M.

Noted. I havent seen 2800 trdrd 4. Highest i saw was 27xx

[cadaveca]

http://www.ap.com/


A dedicated audio analyzer from those guys is what one needs to measure this stuff properly - unfortunately it falls beyond the scope of most can use etc. Thats what we use at HQ to measure our stuff for the SNR figures in the marketing materials.

RMAA loopback testing is limited by the input SNR/dynamic range as well - so there will always be offset between the actual SNR/DR and what one can get with RMAA. The guides on RMAA are written so one can set the software up and at least get to the best possible measurement - hardware/software limitations of this form of testing withstanding.

-Raja

Yeah, I didn't mean to say that your reporting is flawed, rather my environment is different. That said, I still manage to replicate audio performance characteristics in a reasonable manner that manages to show the differences between audio designs, and was more reflecting on my "test equipment" than your testing methods. The given guide mentions the basics, cable used, disabling audio "enhancing" software, etc, but I built my own cable and use far different settings in RMAA than you guys suggest. It's not a big deal, at all, in my books, but I do feel the distinction has to be made that what I do, and you guys suggest, are different.


And the HERO audio stands out, big time, in my testing, and you guys deserve some congratulations on a job well done there.

[Raja@ASUS]

I understand.

Formula is another level up this round on the audio side as well.

[RPGWiZaRD]

I understand.

Formula is another level up this round on the audio side as well.

Please, can you confirm whether Gene and Hero uses exactly same configs, it mentions 115dB SNR on Gene but says nada about Hero why I get a little suspicious. I hope noise-wise they are pretty the same as I'd prefer Hero slightly if both was the same but if Gene was even slightly better (even say 1dB lower THD Noise for example), I'd go for Gene... :P Don't expect you to know about this but if you happen to know, then please share as then I might be able to decide which one to purchase sooner.

[Anusha]

so are you guys saying i can get rid of the the X-Fi Titanium card and feel no difference in audio quality if I go for the Maximus VI Hero? (btw, I am no audiophile)

[djt]

Question for Raja@ASUS: you mentioned on page 10 of this thread that if you are if you are using only two memory modules use the red slots which does go against what the manual for the Maximus VI Extreme says.

Is that what you recommend regardless of the memory speed or module size?