just got a brand new 800W Titanium Certification PSU like 2 weeks ago. 1 x 8 / 4-Pin EPS / ATX 12V connector
No dual 4 pins?
i think they're dual pins, so just like 1 4 pin in each 8 pin?
No your psu has a single 8 pin, which should be pluged into the main 8 pin cpu power socket.
The additional 8 pin cpu power socked, "can" be used if you have a second 8 pin cpu powerplug,
or if you have 2x 4 pin (dual 4 pin) cpu powerplugs additionally.
I remember hearing in one of the computex videos (dunno which one of the 100 hundred videos) that there was a requirement for at least 8 pin EPS and a 4 EPS, and that makers were just going with dual 8's anyway. But it did sound like you needed the 8 and 4, unlike on some X99 board's where the second plug was only there for extreme overclocking. Might be wrong though.
1 Like
Yeah that is something we have to wait and see for.
I have not checked out manny X399 boards yet.
The Aorus X399 Gaming 7 has an 8 + 4 pin indeed.
But most boards shown at computex are in a pre state.
So there might still be some things changing to them.
Like the Aorus X399 Gaming 7 board was only a 6 phase as it was showing.
But wenn its released Gigabyte is going to add it up to 8 phases.
With IR3553 powerstages.
1 Like
Very true and definitely a good note to keep in mind. But, with that said, there wasn't any boards shown with anything less than a 8 + 4 pin configuration. Admittedly there are also no lower-end board's being shown, but to this point we've not seen any single 8 pin boards.
Yeah i think that most thread ripper boards are going to be pretty highend.
The additional 4 pin might or might not be needed, i dont have much info on that yet.
1 Like
I thought it was pretty much common knowledge that the 7700k edges out the 1700 in fps. Were being pedantic now guys. I'm not saying "the 7700k trashes the 1700 in gaming". I'm saying its a "small hit". In most cases it's entirely unnoticeable. Which I never said it was to begin with.
This is good for them it will teach them a lesson that even the highest up on the totem pole are not untouchable
1 Like
I do have some.. it's unlikely you will see very many small boards out of this chip. The simple fact is with 64 pcie lanes and the kind of stuff attached is going to make the board difficult to shrink. Plus the power delivery and even if you shrunk the ram slots down from 8 to 4 it's still hard to make microatx which is the smallest I see the boards going.. sorry even if they did make a mini itx it would probably be a narrow Ilm cooler if that because of the heat it's unlikely any narrow Ilm would cool it so seeing that fact probably not mini itx
Sorry mate, not sure where your info is coming from, based on everything I have seen, the dual socket epyc board is 128pcie lanes 64 for cross cpu comms/infinity fabric with the other 64 for board comms. Also 8 core ccx's? I have not seen a pic of the actual silicon but the epyc 32 core is 4 dies on one SoC, so ok, it may be.
You may be right, zen may only cut consumption by 33%, time will tell.
IPC, more sockets, same core count, at what cost? competition is not just ipc, or even core count, its cost of operation and price of the product, Intel can have a 7% ipc advantage, but the 28 core skylake Xeon Platinum 8180 will be $12,800 each (https://en.wikipedia.org/wiki/List_of_Intel_Xeon_microprocessors#.22Skylake-H.22_.2814_nm.29) so ya lets see what epyc sells for I am pretty sure it won't be 12k per SoC, but again time will tell.
I mean shit you wanna talk competitive... I can get a powerpc machine set up for LESS than a skylake machine and still have the same processing power 
But then again arguing decimal numbers of metrics is rather stupid isn't it?
2 Likes
Skylake-EP isn't formally released... That wikipedia page is still rumor mill stuff from what I can see. Intel hasn't listed the Skylake-EP chips on ARK and we haven't seen a formal release either. That price is unconfirmed and for a part that hasn't even been formally shown by intel.
Your just making up numbers mate.
Ryzen is power efficient yes, but its not 50% better, it's not 33% better, its much, much less than that. Comparing the 6900k to a 1800x is a 5 watt load difference. That's also not intel's newest architecture. TDP is NOT power consumption, it stands for "Thermal Design Power" which is NOT the same as power consumption.
I expect that all boards will be full sized.
well then, how do you know that it will be competitive if its all just rumour, you can not have it both ways, either you compare what is in the market now, or you are just blowing wind, mate.
Its ok, so are you, but that is beside the point, my original post was talking about a current product in the market vs epyc which is not, it was actually a what if or a thought piece about the possibilities that are coming, xeon skylake may kick epyc's butt, I really don't know, and as you stated above you don't either. I was postulating a roadmap which appears to have completely disrupted the entire cpu space from consumer to workstation to server farms and what that may look like moving forward. If you think intel is not nervous and in near panic mode over what is coming out of AMD, you have that right, but it is all just opinion, my opinion is Intel got caught out, unaware and is now trying to figure out what the heck to do, thats all. Meanwhile AMD is executing a roadmap with some bumps and hurdles but none the less they are executing and its shaking things up. EOM.
Because I'm not actually making up numbers? We know Skylake-EP will be 32 core max, that's known (for more than a year now there have been leaks saying 32 core skylake-EP is coming, and seeing a broadwell-ep was 24 max its both reasonable and now substantiated enough for me to safely say that yeah its happening). We know the IPC of skylake as a microarchitechure is superior to that of Zen as a microarchitechure, that's known. We know intel supports up to 8 socket configurations with their highest end server platform, that's known.
Literally posted a benchmark that showed comparable Zen based and Broadwell based parts and their power usage under a full cpu load, likens it to you showing nothing and claiming 50% power usage reduction. Yeah okay mate that's the same thing.
Intel is still selling consumer CPU's at $2K for 18 core. It is still a vaporware product that does not exist. If threadripper comes out at half the price or the rumored 800-900USD it will smash intel once again. Intel will have single thread performance but it is a multithreaded world.
yep, you are right, you posted a chart showing the 1800x and a 6900k with power usage as 5 watts different in an actual cinebench15 test. so here is the deal, neither of those parts are xeon or epyc, I hope we can agree on that. those parts are both consumer parts and are both in the market right now so based on that, the facts are under that particular load, cinebench15, the 6900k uses 160.5 watts of power at load and 70.2 at idle where the 1800x uses 155.1 at load and 37.6 at idle. So ya in that test for what ever time that test takes to run there was a 5 watt difference at load and a 33 watt difference at idle. That benchmark/test/graph really has nothing to do with what epyc's tdp or skylake's tdp or actual power consumption will be.
Ok, there is a "leak" that skylake has a 32 core cpu, last time I looked "leaks" do not equate to factual statements made by companies. So while you may "know" this is coming, you said it was a "leak".
Really we "know" skylake is superior?
so, the 1700x beats kabylake according to these tests in IPC, kabylake and skylake are basically identical in IPC so sorry mate, skylake does not beat like for like infact kabylake is 13 behind when both are at the same 3.8 ghz speed. But i will just say according to this, the 1700x and the kabylake are both at 111 stock, so no, skylake does not beat the zen architecture at ipc in actual testing according to this particular set of tests published on this particular website.
So again, i would like to point out my original post, was a thought pc about what may be going to come about, you attacked my assumptions and opinions, no problem, until we get epyc and skylake-ep actually released into the wild for testing neither of us really has anymore than speculation or rumor or supposed known facts from "leaks" to work with. Comparing consumer intel parts to consumer ryzen parts is great, intel wins some stuff ryzen wins some stuff. Take your pick of which one you want to spend your money on based on your usage scenario. Or what ever makes you decide to buy one product over another.
In conclusion, skylake does not beat ryzen in ipc, skylake does not beat ryzen in power usage, skylake does not beat ryzen in pcie lanes and most of all skylake does not even come close to beating ryzen in value. Have a nice day.
Do you not understand that it's not a benchmark to show the EXACT difference, but one to show that its really not a 50% difference or a 33% difference? Its a demonstration of 2 comparable 8 core parts from AMD and Intel that show the rough difference in TDP, so we can get a rough sense of the power consumption comparison of the two architectures? Its not unreasonable to extrapolate that from that we can see that there is a slight advantage for a Zen core over a Broadwell core.
Engineering samples of the chip have been seen in the wild, there are postings on benchmark sites of 32 core intel engineering sample chips, and a 32 core chip is a linear, reasonable, logical step-up from the current 24 core maximum. Yes intel hasn't had their official press release of the chip, but we basically know that yeah there is going to be a 32 core chip.
Congrats, in that one benchmark (where you conviently left out the text UNDER the benchmarks) AMD has a higher IPC. This is the text that was under that:
It’s still very important to note that one benchmark isn’t the be all and end all of performance and IPC.
It was also noted that in the Benchmark you encluded, the intel kaby lake cpu had a distinct cache disadvantage and that the result was much closer (still in ryzen's favor, but this can obviously be seen as an outlying test) when a cache-comparable cpu was used.
It's commonly accepted that Zen architecture as a whole is roughly the same as broadwell, and frankly there are a million other tests that show the opposite from the one you provided.
single thread performance shown in cinebench r15
multi-thread parity shown in handbreak
multi-thread parity shown again in cinebench r15 multi-thread