|
Basically wait until Jan5-8th.
|
#
?
Dec 31, 2016 00:47
|
|
|
|
| # ? May 21, 2024 13:21 |
|
|
Not even then. I mean, we're somewhat likely to know *more*, but I would literally wait until the 17th.
|
|
#
?
Dec 31, 2016 01:37
|
|
|
That's what I was planning on.
|
|
#
?
Dec 31, 2016 02:49
|
|
|
I predict the Zen SKUs we want to buy on a not-retarded pricing will come by Q2 2017 earliest, most likely in Q3. I'm actually hoping Zen does well enough, but also not so well to the point that AMD feels compelled to price match Intel on the same core/thread config and also follow the K/non-K CPU + Z/non-Z chipset crap.
Palladium fucked around with this message at 05:34 on Dec 31, 2016 |
|
#
?
Dec 31, 2016 05:31
|
|
|
So, um. Heavily salt this poo poo. https://www.reddit.com/r/Amd/comments/5krghq/remember_the_canard_pc_magazine_about_zen_it_has/ In short: CanardPC (those French guys who got their hands on an engineering sample Zen chip) embedded an easter egg into their article. The easter egg says that Zen can clock to 5 GHz on air. CanardPC have now come clean with additional details: http://www.cpchardware.com/cpc-hardware-n31-precisions-elucubrations/ Rough translation, parenthesis are my commentary: * We wanted to save something as an exclusive for our own publication in case our graphs got reposted to hell BY OUR COMPETITORS! The hell's up with that, guys? Ever hear of something called journalistic integrity? * Hiding easter eggs is nothing new for us. * Comparisons from the original tests are based on single-core performance with multithreading enabled. * Our contact would not let us publish raw numbers because if we did, it would have been very easy for AMD to figure out the combination of engineering sample CPU, BIOS, AGESA, etc and compromise the source. * They did not do the overclocking themselves. * Air cooler used was something huge. The word used is "imposant", imposing. (I wonder, does Prolimatech still make the Megahalems?) * Bit of a pullback on the 5GHz claim. Near the end of the article they say the chip only "froler" (trans: brushes) with 5 GHz. (I interpret this to mean they're within 'palm on the wall to aim down into the bowl first thing in the morning after waking up'-pissing distance of 5 GHz, and expect the end product to hit 5 GHz based on this still being engineering sample silicon.) * Multiplier not locked and comes in steps of .25x * VRMs on the sample motherboard couldn't keep up these results across all cores, will need much beefier power-delivery to do it proper. (Gigabyte? I'm looking at you) * Other Zen engineering samples have been distributed to overclockers, we may see OC demos at CES. * Be Sure To Drink Your And that's about as much good as my four years of middle school and high school French are worth. SwissArmyDruid fucked around with this message at 14:35 on Dec 31, 2016 |
|
#
?
Dec 31, 2016 09:48
|
|
|
What a fun role reversal it would be if Zen ends up beating newer Intel hardware at matching product segments by pure clock speed brute force. Up is down, hate is love, and old Netburst engineers are hitting the whiskey. (Well, not really, I think back when it was brand new Intel was hoping Netburst would be cracking 6+ GHz in shitbox Compaqs at some point in the mid oughts.)
|
|
#
?
Dec 31, 2016 11:02
|
|
|
20GHz single core performance at 600W TDP would be cool and good 
|
|
#
?
Dec 31, 2016 11:52
|
|
I CANNOT EJACULATE WITHOUT SEEING NATIVE AMERICANS BRUTALISED!
|
blowfish posted:20GHz single core performance at 600W TDP would be cool and good At 20ghz the fundamental issues speed of light signal propagation start to become a serious issue at the 1-2cm level. Even if you had arbitrarily fast switching transistors, light can only go about 1.5 cm in 1/20 billionth of a second. The only way you could get around that would be with some sort of bullshit tier 3d lithography process that manages to wedge all the L1/2/3/4 cache on one level and all the logic gates on top of it on another level, and even then you run into issues with coherency.
|
|
#
?
Dec 31, 2016 12:10
|
|
|
Also, https://twitter.com/InstLatX64/status/813539775944785921
|
|
#
?
Dec 31, 2016 12:20
|
|
|
So AMD fixed their cache problems it seems. I still want to know what fullblown excavator (8 modules, L3$ and latencies worth a drat) would behave like. It seems like even if everything went perfectly AMD would have only competed with Sandy.
|
|
#
?
Dec 31, 2016 12:27
|
|
|
Methylethylaldehyde posted:At 20ghz the fundamental issues speed of light signal propagation start to become a serious issue at the 1-2cm level. Even if you had arbitrarily fast switching transistors, light can only go about 1.5 cm in 1/20 billionth of a second. You don't engineer around it, you market around it. Just keep some combination of small things in it at x*y=20GHz and build the rest of the chip around that to mask the physical limitations. Continue to market as 20 GHz chip, win market share because managers want crimson databases that go fast because the red ones go faster sort of effect.
|
|
#
?
Dec 31, 2016 12:37
|
|
|
Hahahah holy poo poo, everything is about to change.
|
|
#
?
Dec 31, 2016 16:51
|
|
|
If it hits 5Ghz I'd pay as much as $130 for it. Otherwise I'll stick with my 2500k.
|
|
#
?
Dec 31, 2016 17:08
|
|
|
Pryor on Fire posted:Hahahah holy poo poo, everything is about to change. Just like the last 3 times everything was about to change because of AMD. I have absolutely no confidence in their benchmarks.
|
|
#
?
Dec 31, 2016 18:19
|
|
|
ItBurns posted:If it hits 5Ghz I'd pay as much as $130 for it. Otherwise I'll stick with my 2500k. You should, if used 2500K and P67/Z68 boards are still selling for ~$80 and ~$100 after Zen. People love to pay retarded amounts for old PC gear for some odd reason instead just buying better brand new parts that are only slightly more expensive, and has been that way since forever. Heck, 3770K are going for $280 on Ebay FFS.
|
|
#
?
Dec 31, 2016 18:38
|
|
|
I'll loving believe it hits 5 gigawiggles when I see it somewhere reputable. Otherwise yeah loving right.
|
|
#
?
Dec 31, 2016 23:11
|
|
|
Wistful of Dollars posted:I'll loving believe it hits 5 gigawiggles when I see it somewhere reputable. Otherwise yeah loving right. Canard PC is reputable, just that the 5Ghz has the cavaet of being achieved only on one core, and that the board (of unknown quality) had poor VRMs to try pushing it any further on anymore cores.
|
|
#
?
Dec 31, 2016 23:51
|
|
|
Ebay sellers always try to rob you blind on computer parts, though. If you want a more realistic price estimate, go on fs/ft hardware forums. e: after a quick glance, 3770Ks can be had for about $200, which is still too much but definitely an improvement over $280 lDDQD fucked around with this message at 00:01 on Jan 1, 2017 |
|
#
?
Dec 31, 2016 23:55
|
|
|
drat, a bunch of 3770ks went for 280ish recently though. Had to check the completed and sold listings. I suppose the people buying these were totally adamant about not getting on the chain upgrade train. You know. Where you need a new cpu, but its a different socket type so you need a new motherboard and ack. DDR3 wont work on it so add another 2x sticks of RAM and oh gently caress it, my PSU is old so I might as well just build a new PC.
|
|
#
?
Jan 1, 2017 01:21
|
|
|
Yeah its great I just sell all my parts everytime there's even an incremental upgrade cause resale is so ridiculous so I always have the best poo poo and nothing ever breaks cause its always new
|
|
#
?
Jan 1, 2017 02:29
|
|
|
Or someone with an iMac who likes the rest of it but wants hyperthreading.
|
|
#
?
Jan 1, 2017 02:30
|
|
|
WanderingKid posted:drat, a bunch of 3770ks went for 280ish recently though. Had to check the completed and sold listings. Yikes. I've bought whole used Dells with 3770s in them for less.
|
|
#
?
Jan 1, 2017 15:56
|
|
|
Chipzilla starting to maneuver. Kaby Lake-H got axed. http://www.guru3d.com/news-story/intel-cancles-kaby-lake-h-kaby-lake-refresh-will-be-core-ix-8000-series.html
|
|
#
?
Jan 2, 2017 13:50
|
|
|
But could they possibly have a way to counter the wildest card of all? Jim motherfucking Keller-san??? This is very exciting.
|
|
#
?
Jan 2, 2017 14:07
|
|
|
Don't read too much into it, -H parts are mobile parts, and they got axed because it's essentially the same Broadwell part on the same process. I do like to think that if not for Zen, Intel would have just kept that little boondoggle.
|
|
#
?
Jan 2, 2017 14:36
|
|
|
A comment on the freesync 2 article had mentioned that this "automatic HDR rendering" and TrueAudio Next are the only two unused Polaris features left. Will anything happen with TrueAudio Next? It doesn't seem to see a lot of adoption. That said we did just see Valve push out a HRTF sound option for CS:GO.
|
|
#
?
Jan 3, 2017 03:48
|
|
|
SwissArmyDruid posted:Don't read too much into it, -H parts are mobile parts, and they got axed because it's essentially the same Broadwell part on the same process. s/Broadwell/Skylake Agreed that anyone who has been paying attention should know that Kaby Lake is basically Devil's Canyon 2 though - it got promoted to "optimization" because Intel doesn't think 2 year turnaround on new processes is feasible anymore, not because it's a big deal.
|
|
#
?
Jan 3, 2017 16:29
|
|
|
Pryor on Fire posted:Hahahah holy poo poo, everything is about to change. For the way they were talking about their new gpu earlier this year, I was expecting something huge of this generation to at least match a 1070. It didn't even come close; it matched the past generation 970. It's ok to be hype about something but wait until something hit the market to judge it. edit: rephrased my argument Bleusilences fucked around with this message at 18:19 on Jan 4, 2017 |
|
#
?
Jan 4, 2017 00:03
|
|
|
Thought just occurred to me, but aren't cat cores super cheap to produce, and there will be AM4 board with no chipset (A320 and A300)? Really hoping for a return of cat cores, maybe with some upgrades, as AMDs new lowest of low end and a full replacement of AM1 as it gets folded into AM4. They already have them on 14/16nm process, would be cool to get 2/4/6/8 cat cores with 128/192/256/320/384 GCN4/5 cores. I think there is space for them and with some improvements to L2, addition of small L3 and moving to dual channel they'd be a better low end alternative to construction cores. As is, AMD's 2017 line up seems to be 8C/16T Black - 400-499$ 8C/16T - 300-399$ 4C/8T Black - 200-299$ 4C/8T and 4C/8T APU - 150-249$ 2C/4T APU - 75-149$ ????? 20-100$ Based on prices of 28nm Kabini, they seem like a perfect fit to match Celerons and Pentiums (by leveraging more cores and better iGPU). As it is, the Excavator offerings seem to be OEM focused and have no leverage over Celerons and Pentiums offered by Intel. Also, high end AM4 MSI board seen 
EmpyreanFlux fucked around with this message at 15:51 on Jan 4, 2017 |
|
#
?
Jan 4, 2017 15:30
|
|
|
So much wasted space around that socket, man. I feel like if they didn't need to have those plastic hooks, you could bracket the socket with RAM slots as with Intel.
|
|
#
?
Jan 4, 2017 16:32
|
|
|
I think it's to make space for the industrial warehouse cooler you will need for overclocking.
|
|
#
?
Jan 4, 2017 16:37
|
|
|
Har har, yes, that's right, because AMD = space heater. Come on, dude, that poo poo's unscientific. Learn to abandon your biases and take a fresh perspective on the new design.
|
|
#
?
Jan 4, 2017 16:42
|
|
|
What's unscientific about needing a big cooler for overclocking?
champagne posting fucked around with this message at 17:14 on Jan 4, 2017 |
|
#
?
Jan 4, 2017 16:43
|
|
|
Boiled Water posted:What's unscientific about wanting a big cooler for overclocking? The difference between "want and "need", as per your posting.
|
|
#
?
Jan 4, 2017 16:54
|
|
|
Glad to see a U.2. port on the motherboard - if they can manage to create a high end mini-itx board with U.2. and an one or two M.2. ports, i'm sold (if the CPUs don't stuck that is).
|
|
#
?
Jan 4, 2017 17:10
|
|
|
In retrospect, it was probably a reference to "5 GHz on air". My apologies.
|
|
#
?
Jan 4, 2017 17:14
|
|
|
quite alright
|
|
#
?
Jan 4, 2017 17:15
|
|
|
Deathreaper posted:Glad to see a U.2. port on the motherboard Wait wh--HOLY poo poo  u.2 is a goood future -- if we can embrace it.
|
|
#
?
Jan 4, 2017 17:40
|
|
|
Oh man, who was it saying Kaby Lake was going to come out and crush Xen? http://arstechnica.com/gadgets/2017/01/intel-core-i7-7700k-kaby-lake-review/ I love the tomb raider benchmark where the 6700k @ 4.2ghz beats the 7700k @ 4.5ghz in FPS
|
|
#
?
Jan 4, 2017 18:49
|
|
|
|
| # ? May 21, 2024 13:21 |
|
|
...the gently caress? Either Intel is having some SERIOUS foundry problems with 10nm, or they're just throwing this generation so they can have competition and dodge any SEC investigations. How do you actually screw that up?
|
|
#
?
Jan 4, 2017 19:59
|
|