ASUS P8Z77-V DELUXE DDR3 T-Topology overview

Moogle Stiltzki

Junior Member
May 3, 2012
4
0
71
This article basically tells you what it does (pics illustrations included)
http://rog.asus.com/101102012/maxim...r3-t-topology-design-on-z77-h77-motherboards/


And Sin822 made some further feedback regarding this feature which Anandtech was the first to point out.

Let's be VERY clear, this does NOTHING to improve clock for clock performance, it doesn't make the memory subsystem faster, it just helps OC ONLY when ALL DIMMs are filled up.

If you run the memory at 1600mhz it will run with the same performance at 1600mhz on another board. This technology is saying it can get memory upto 15% more mhz, of course only with 4DIMMs populated(as it only applies to 4-DIMMs). Bsically a channel is directly hooked up to one DIMM, that is why you skip one slot when installing 2 DIMMs. Now instead of having the second DIMM right next to the one that is directly connected to the IMC be connected to the first DIMM, it has all 4 DIMMs connected to the IMC, that isn't a breakthrough. lol.

SO this is JUST for showing off, just like I showed off 3GHz with 4-DIMMs with 16GB! lol, it has no practical benching uses b/c you would never bench with all 4DIMMs(the probability of having one stick OC less than the others increases as you increase memory stick count). There is also no practical uses for OCing memory for a 24/7 RIG or anything really especially not over 2800mhz which is easy to do on Z77 and IB, easy. Plus they don't even make memory that OCes that high, highest rated will be 2800mhz from team group and g,skill the memory i posted wasn't some random kit either.

So GIGABYTE boards can do 3GHz on the memory with 4DIMMs on Z77 with 16GB already proven, more than enough and over the top considering the best memory g,skill has doesn't goto me. I also sure ASUS will do better, for god sakes coolaler beat my score and he used the same board I did! hahahahahahaah

This is my source, Raja@ASUS: http://www.xtremesystems.org/forums/showthread.php?280224-Asus-Ivy-memory-treat backing everything i say pertaining to what this tech affects. At least he isn't making it sound like the best thing since sliced bread like andatech makes it seem.

So this technology doesn't have ANY impact when using dual channel with 2-DIMMs, it only helps when you add an extra stick to each channel, it is cool, but it also doesn't mean other companies don't already do this or something else to increase clocks

IMO this type of information, and vague placements in adnatech articles is not only deliberate, and for some reason i am not allowed to talk about the economics behind this, but let's just say when you hear it increase "performance" you think actually clock for clock memory, however that isn't true, they should say increase mhz starting at 3GHz. As that is where it would have to start as GB boards can do 3GHz.

Also if this makes Intel Engineer blsuh, i mean come on Intel you are just going to let them slander you like that?, if you make them blush enough, when you do something totally different, you are then talking about a HUGE delay in product launch as Intel will want to put it through a year of QA testing and then call it isn't own technology as that is what the partner agreement states. It isn't worth it to go beyond and above Intel spec in things like this as you wont make money off of it because your competitors will have it.. Like if ASUS was able to let's say get quad channel working on Z77, which they aren't, i am just saying, then Intel would have to provide that technology to other partners.

This t-crap is only marketing which sounds really technical, but really isn't and doesn't have any impact on 99.999999% of people, even Ln2 OCers will see no benefit from this b/c only some benchmarks like fast memory, and to run memory that cast you are going to lose out on timings. It also doesn't help for WRs for memory peed b/c llano still has better IMC, also you don't break memory WR on 4-DIMMs, you use a single DIMM lol.

Source:
http://www.overclock.net/t/1241807/asus-t-topology


In summary he concludes it's not such a big deal, and that it would require that all 4 dimm slots be occupied to be used.


I wonder what Anand has to comment regarding these 2 new sources of info i wonder :hmm:
 
Last edited:

Moogle Stiltzki

Junior Member
May 3, 2012
4
0
71
Oh and Raja@Asus hit back with this comment

The decision to implement T-Branch was made because most people are now running 4 DIMMs (prices are very reasonable). It provides a little more system overhead, by reducing slot to slot skew. Any skew has to be countered in the IMC read/write schedule and eventually eats into margin. So yes for 4 DIMMs only, and also yes that its for the guys that like to push. The reason Intel was said to be impressed is because ASUS had to model the transmission line impedances and layout themselves - something most vendors won't and do not do. Most boards use the Intel generic layout and transmission line impedances so there is some truth behind the part that Intel were glad we pushed the envelope of their design.

-Raja

Edit: one thing I should add that may be of benefit to those pushing 4 DIMMs, you should find that T-Branch may allow lower operating voltages for the memory bus (VDD) against conventional topologies. Any variance in skew between DQ and DQS lines due to daisy chaining of the single ended DQ lines, will have to be countered by fiddling with VDD to maintain a slew rate that keeps the DQ signal well centred with the DQS.

No matter what Sin says, we put a huge amount of work into this. We did not use Intel's stock layout and modelled the layout to ensure that we extended the 4 DIMM OC margins of the platform. All Z77 boards get T-Branch while the ROG boards are optimized further still. We don't just change VRMs on boards and leave it at that. Like all vendors we do have time constraints to work to, but we do have resources at our disposal that allow us to cram more of this kind of work into our designs.

The funny thing about Sin's post is that he does not even know what T-Branch is.

Source:
http://www.overclock.net/t/1241807/asus-t-topology


Hm...
 

Moogle Stiltzki

Junior Member
May 3, 2012
4
0
71
Sin later came back with this

Once again this is a forum people can spell however they want as long as it is coherent. Thomas Jefferson once said "I have nothing but contempt for anyone who can spell a word only one way." which means you can never trust a man who can spell a word only one way. This is a forum not a thesis paper, Anandatech is not a real word anyway( I have to look up how to spell it, but from now on i will call it AT in posts after this).

For all of you who think I am some biased SOB as Raja and Bingo13(who is Raja's boss at ASUS), here is the proof that i am not:
. I don't ever use the search function, but now that I see it works let me show you guys something, okay?

Here is where I told a guy and instead of telling him to buy the G1 Sniper 2 i told him to get a Maximus 4 Gene and a sound card, and it looks like he did: http://www.overclock.net/t/1186098/dated-i-know-looking-for-help-on-g1-sniper

Here is another thread in which I not only recommend the M4G and M4E, i go on to state the board's benefits over the others, and I defended asus to the dime, however i of course get no credit for it. i make many posts in this thread all of which help ASUS VERY much: http://www.overclock.net/t/1186516/lga1155-mobo-advise-evga-or-asus

Here is an example where I recommend both GB and ASUS, and i do this very often:
http://www.overclock.net/t/1102858/asus-maximus-iv-extreme-z-or-gigabyte-ga-z68x-ud7-b3

See guys I am not some biased POS. Please have you ever seen el gappo or praz recommend GIGABYTE? Or have you ever seen Raja or Bing13 do the same? Please stop attacking me, you can see that I can be objective, so please let me be. Hey praz nice to see that the bat signal got you!!!!

listen guys I am not against this product development, i have never been against ASUS's Z77 lineup, I don't care if you buy GB or ASUS, i dont' get paid nor do i get more samples if I do. My only issue is with the way ASUS has been marketing things lately, they have been very vague with how this works and what it benefits, thus to mislead users to believe that their products are the only way to go. IMO i have recommended ASUS many times here.

Another point is that SPI sure might like high frequency but it also demands you have very good timings, and there is no 4 dimm memory kit out there that is going to give you the timings some of the good dual channel kits can do at 2400-2700 which would be the optimal benching range. You show a 300ms difference, how what else can make that 300ms? How about the different efficiency of the different users? Cannot one user be more efficient? How about higher memory frequency like a few mhz b/c of the fact that two of those modules will OC higher than the other two? That is where my whole issue with the marketing comes into play, the simple logic of the matter is that this has no benefit to the majority of users(including benchers) as if any Z77 board is designed correctly it should be able to do 2400-2700+MHz stable bench-able with 4 DIMMs, however ASUS's tech can take that number higher, I can believe that and agree with it, but is it practical for benchers? The practicality of it I have not seen, yet.


Also El Gappo, my beef with andatech goes back a long time from before i review GB boards, their site has posted false information about my favorite SSD and I was extremely angry that they didn't correct themselves. The author was the owner of the site of which he named it after.


Also El Gappo i didn't ask anyone to delete my posts or Rajas or Garys and I didn't ask anyone to lock the thread or try to mediate. The issue is plain and simple that ASUS believes that I am a grave and imminent thread to ASUS, and they are taking care of it, which I am sure you are well aware of as they have sent out the bat signal to all the ASUS sponsored guys.


You guys are saying I don't understand, however this is what I know:
I know that this technology only helps 4 DIMM overclocking
I know that this technology doesn't offer clock for clock gains in efficiency
I know that this technology helps OC 4 DIMMs above and byond 2800MHz

Now here is what I know about OCing and Ivy:
Ivy Bridge doesn't have great IMCs in all CPUs, not many will do >2800mhz
Other Z77 non-ASUS boards can bench 4 DIMMs at 2800mhz when the memory is properly optimized.
Most people have already proven that they bench in SPI not with over 3GHz, but instead where timings and mHz balance, and the reason I hit at the marketing here is because I think this technology only helps at speed beyond what is currently the optimal range.

The technology is awesome, however it's usefulness is beyond what can be done with current DRAM modules and that is why I have issue with the scale at which it is advertised. That is just my 2cents, if you have any issues please tell me what I am wrong about. Of course I guess now it makes sense to blow up things like this, after all manufactures blow up phase count when all you need is 4-6 high quality phases. I guess it is part of the game and I should get used to it.

The bottom line is that I do a lot in this community to help users, I give them advise which i deem fit from my own use of the boards. if I have never used a board I will say I have never used it, however I have used Z68 boards from all manufacturers, and I decided to recommend the M4G, so why is ASUS so uptight about?

Now if i am wrong about what I said about T-Toplogy and how it works please let me know exactly how i am wrong and prove to me that i am wrong, don't just blatantly state that I have no idea what i am talking about and then not correct what i said or explain it.

Source:
http://www.overclock.net/t/1241807/asus-t-topology/20


But looking at their arguements, my conclusion as newbie says essentially if your not an overclocker it's like meh :X

If i'm wrong on this let me know :X
 

Moogle Stiltzki

Junior Member
May 3, 2012
4
0
71
picture courtesy of Raja@asus

600x450px-LL-8828e27a_memory1.jpeg


600x450px-LL-55d58b03_memory2.jpeg
 
Last edited:

ViRGE

Elite Member, Moderator Emeritus
Oct 9, 1999
31,516
167
106
A link would have been sufficient ..
Indeed.

I'm still not sure I understand why you posted this, Moogle. Is there a question in here somewhere? Or are you just trying to draw more people into this?