|
Holy poop.
|
# ? Jun 19, 2017 15:16 |
|
|
# ? May 24, 2024 19:41 |
|
I'm just going through the Anandtech review, and there's already the first "minor" difference between the SKUs. The 10C has two execution ports for AVX-512 FMA ops, while the 8C and 6C only have one. So there's already going to be instruction level differences in some operations.
|
# ? Jun 19, 2017 15:26 |
|
Combat Pretzel posted:I'm just going through the Anandtech review, and there's already the first "minor" difference between the SKUs. The 10C has two execution ports for AVX-512 FMA ops, while the 8C and 6C only have one. So there's already going to be instruction level differences in some operations. Does the concept of LCC, MCC, HCC dies still exist with this new mesh layout? Is it as simple as the 8 and fewer core CPUs being LCC dies, and 10-14 being MCC? Edit: Whoa, if X299 mobos weren't so expensive then the 7800K looking like a fine value.
|
# ? Jun 19, 2017 15:31 |
|
Twerk from Home posted:Edit: Whoa, if X299 mobos weren't so expensive then the 7800K looking like a fine value. Seriously, these motherboard prices are crazy. Asus's cheapest X299 board costs more than the Crosshair VI, their most expensive AM4 board repiv fucked around with this message at 15:41 on Jun 19, 2017 |
# ? Jun 19, 2017 15:36 |
|
power consumption seems very inconsistent across the reviews, i hope that's an indicator that it's still being fine tuned. the pcperspective review shows only 9W higher idle power consumption than a 7700K, although all their ryzen idle numbers are much lower than what I've seen from other reviews. e: Oh man, 7900X draws 70 watts more power than the 6950X and both have a 140 watts TDP eames fucked around with this message at 15:48 on Jun 19, 2017 |
# ? Jun 19, 2017 15:40 |
|
repiv posted:Seriously, these motherboard prices are crazy. Asus's cheapest X299 board costs more than the Crosshair VI, their most expensive AM4 board
|
# ? Jun 19, 2017 15:41 |
|
I wonder if board prices will calm down if the different board partners roll out separate "Skylake-X Only" vs "Kaby lake-X only" versions which don't have to deal with the complicated bullshit of differentiating between on chip voltage regulation vs off chip, 2 vs 4 channel ram, different PCIe configs, etc. Or if they're prohibited from doing things like that by Intel terms which say anything using an X299 chipset has to support both stupid CPU lines on the same board. It really seems like they're adding a shitload of complexity for no real gain.
|
# ? Jun 19, 2017 15:42 |
|
Where are you guys getting the prices for the X299 boards? I'm trying to scope out a motherboard to get in the next following month
|
# ? Jun 19, 2017 15:52 |
|
Rabid Snake posted:Where are you guys getting the prices for the X299 boards? I'm trying to scope out a motherboard to get in the next following month https://www.bhphotovideo.com/c/search?atclk=fct_a_cpu-socket_3756-lga-2066&ci=19864&fct=fct_a_cpu-socket_3756%7clga-2066&N=3835434462 Looks like a range of 250-489 here for pre-orders. Newegg has 229-499$ https://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100007627%20601299335&IsNodeId=1&bop=And&Order=PRICE&PageSize=36 The cheapest X99 boards look more like 180$ minimum, so that's a 40$ jump in board price. Gwaihir fucked around with this message at 16:03 on Jun 19, 2017 |
# ? Jun 19, 2017 16:00 |
|
Done with the Anandtech review. Too bad they skipped the gaming benchmarks due to issues. Threadripper reviews and benchmarks can't come soon enough.
|
# ? Jun 19, 2017 16:15 |
|
X299 board prices seem in line with x99 with asrock leading the pack for value price.
|
# ? Jun 19, 2017 16:24 |
|
anandtech posted:When we started testing for this review, the main instructions we were given was that when changing between Skylake-X and Kaby Lake-X processors, be sure to remove AC power and hold the reset BIOS button for 30 seconds. This comes down to an issue with supporting both sets of CPUs at once: Skylake-X features some form of integrated voltage regulator (somewhat like the FIVR on Broadwell), whereas Kaby Lake-X is more motherboard controlled. As a result, some of the voltages going in to the CPU, if configured incorrectly, can cause damage. This is where I say I broke a CPU: our Kaby Lake-X Core i7 died on the test bed. lovely also this: overclock3d posted:Just as a parting gift, if youve read this far and youve watched the video thank you, if youre here to see if you need to buy the 7820X or for that matter any X299 as the basis for your gaming system we have one thing to say to you. Don't. In the grand scheme of things you either want to go and grab a Z270 /7700K based system or if you want a few more cores for occasional rendering and streaming then just go buy the Ryzen 1700. If you do buy AMD though don't turn into one of the mentalist fanboy keyboard warriors ruining the scene at the moment. but wait there's more! guru3d posted:It has been a bit of a mess with Intel, I can honestly say that. Here in the EU Intel has killed all marketing, press and PR activities. Hence media all over Europe is not receiving any information from Intel anymore, that’s the same samples wise, Intel does not seed samples for review anymore. Hence you’ll see reviews next week based on a handful cherry-picked USA websites that are on a marketing-drip directly from Intel. The rest of the media will need to somehow borrow their samples from Intel partners. Yes, it’s a big freaking mess with everything and anything Intel alright when it comes to media related matters. That said in this preview I wanted to take you guys a little deeper into the architecture and processor series that is Skylake-X, the Core X series. eames fucked around with this message at 16:44 on Jun 19, 2017 |
# ? Jun 19, 2017 16:34 |
|
those thermals Did anyone see my tube of toothpaste, I know I left it over by the Skylake-X assembly line somewhere
|
# ? Jun 19, 2017 18:15 |
|
What the gently caress Did they just not put any sort of TIM between the die and the heat spreader at all?
|
# ? Jun 19, 2017 18:27 |
whats the x axis? load? and why is intel ignoring europa? Are they just that apparently incompetent or are they petulantly spiting them because of EU rulings or whatever?
|
|
# ? Jun 19, 2017 18:33 |
|
Watermelon Daiquiri posted:whats the x axis? load? It's core temperature (red line) vs. temperature of the heat spreader (green line) vs. temperature of the cooling medium (blue line). Basically there's very little heat being actually transferred to the heat spreader, just enough to keep the CPU from killing itself, and then even less heat being transferred to the cooling medium (likely because there isn't enough heat at the heat spreader to be transferred to the cooling medium).
|
# ? Jun 19, 2017 18:56 |
|
Watermelon Daiquiri posted:whats the x axis? load? yeah, power consumption that graph is insane by the way and kind of gives merit to the rumors of a soldered version eames fucked around with this message at 19:09 on Jun 19, 2017 |
# ? Jun 19, 2017 19:03 |
|
eames posted:yeah, power consumption Rumors like this is making me hesitant. I'd be so mad being an early adopter if they rereleased a soldered version a quarter or two down the line.
|
# ? Jun 19, 2017 19:34 |
|
Really looks like Intel should have bumped this launch back, this is a pretty bad debacle. Was looking at preordering somethin but heellllll no to that until they get the BIOS issues fixed. The 7740X getting outperformed fairly significantly by the 7700K is a pretty clear sign that there is a lot of work to be done.
|
# ? Jun 19, 2017 19:56 |
|
Frankenstein's platform
|
# ? Jun 19, 2017 20:00 |
|
eames posted:lovely Hahahaha, what the gently caress is this loving platform, Intel? Some dodgy RAM timings and slightly low stock voltages for full stability on Ryzen look like nothing compared to physical destruction of a CPU just by swapping it in to a board that had another CPU in it before. The TIM situation is a clusterfuck, not much needs to be said about that. The dies are a reasonable size, so the whole "die ain't big enough to solder" argument doesn't work here. HalloKitty fucked around with this message at 20:06 on Jun 19, 2017 |
# ? Jun 19, 2017 20:03 |
|
Inter-core latency looks quite a bit worse too. From PCPer: Almost as bad as Ryzen's latency, but at least it is relatively constant rather than Ryzen's drastic stairstep. Paul MaudDib fucked around with this message at 20:16 on Jun 19, 2017 |
# ? Jun 19, 2017 20:08 |
|
Eh, it's going to take a latency-sensitive process that can effectively leverage >8 threads or memory intensive enough to have to reach through the crossbar to hit pages on a different memory controller before any of that on Ryzen is going to have a meaningful impact. Those things are few and far between, mostly the OS resource scheduler will avoid splitting the process on the crossbar and it won't matter.
|
# ? Jun 19, 2017 20:59 |
|
BangersInMyKnickers posted:Eh, it's going to take a latency-sensitive process that can effectively leverage >8 threads or memory intensive enough to have to reach through the crossbar to hit pages on a different memory controller before any of that on Ryzen is going to have a meaningful impact. Those things are few and far between, mostly the OS resource scheduler will avoid splitting the process on the crossbar and it won't matter. Yeah any massive threaded app has to be numa aware anyway
|
# ? Jun 19, 2017 21:08 |
|
I was nearly sold on a 7820X as I am in dire need of an upgrade in core count, but the thermals on these chips have me really hesitant. I really do not like the idea of delidding a $600 processor. The fact that even with an AIO you are still hitting crazy temps is baffling, Intel must have had a drat good reason to not solder these chips.
|
# ? Jun 19, 2017 21:25 |
|
Yeah, the precious profit line.
|
# ? Jun 19, 2017 21:29 |
|
The HCC CPUs will have to be soldered or they won't be able to hold any decent frequency with air cooling. By the way is it common knowledge that only the 10 core does 2 512bit FMA and the smaller ones only 1? eames fucked around with this message at 21:35 on Jun 19, 2017 |
# ? Jun 19, 2017 21:33 |
|
eames posted:The HCC CPUs will have to be soldered or they won't be able to hold any decent frequency with air cooling. I don't think it was common knowledge before NDA lifted today.
|
# ? Jun 19, 2017 21:47 |
|
Paul MaudDib posted:
What does this chart even mean? Of course the core is going to be hotter than the coolant. How does this compare to [any other CPU]? I guess just showing the temps relative to a 6950x at the same frequency wasn't sensationalist enough. It isn't surprising that if you push a thermally limited (solder or not) chip to its limit it is going to reach the thermal limit. The board frying a CPU because it wasn't unplugged long enough is a waaaaay bigger issue.
|
# ? Jun 19, 2017 22:07 |
|
eames posted:lovely
|
# ? Jun 19, 2017 22:24 |
|
Anime Schoolgirl fucked around with this message at 22:34 on Jun 19, 2017 |
# ? Jun 19, 2017 22:31 |
|
Eyes Only posted:What does this chart even mean? Of course the core is going to be hotter than the coolant. How does this compare to [any other CPU]? If the core is boiling, shouldn't the coolant be a little more... enthusiastic about cooling? Actually grabbing onto some of that heat and doing something with it?
|
# ? Jun 19, 2017 22:33 |
|
Whatever is between that heat spreader I want on my house
|
# ? Jun 19, 2017 22:39 |
|
This might change the comparisons with threadripper a bit. If there's this little thermal headroom it means that it won't be stock vs stock and OC vs OC, it'll be overclocked threadripper against stock sklx. If only AMD could fix their NPT issues and figure out their gcc segfault. The NPT issues alone still have me considering the 7900x. It'd be really nice to see that graph for other cpus as a direct comparison but the absolute values speak for themselves.
|
# ? Jun 19, 2017 22:51 |
|
Is it wrong that seeing these results now makes me want to buy a 1700 non-X and simply OC it to 3.9/4.0?
|
# ? Jun 19, 2017 22:52 |
|
They make me want to wait until generation 2 of Ryzen. Theoretically, I guess I could end up replacing my Intel 2500K with an AMD 2500X. I had been waiting for Skylake-X because I wanted to get more cores and a large bump in single-thread performance at the same time, but these initial reports of insane wattage and temperatures under modest overclocks are not encouraging. A 7820X at stock would still beat my Sandy Bridge's IPC enough to be a clear improvement, but it feels underwhelming and I don't want to notice the increase on my power bill as well if I go higher. I'll be curious to see what kinds of reports come out of the enthusiast community once more people start getting their hands on these and delidding them, though. Eletriarnation fucked around with this message at 23:11 on Jun 19, 2017 |
# ? Jun 19, 2017 23:08 |
|
SourKraut posted:Is it wrong that seeing these results now makes me want to buy a 1700 non-X and simply OC it to 3.9/4.0? No I'd do exactly the same if Ryzen was any good for virtualization. Sadly I don't think that'll happen until the new stepping. Actually I kind of regret not buying a 7700K when it launched.
|
# ? Jun 19, 2017 23:09 |
|
ArgumentatumE.C.T. posted:If the core is boiling, shouldn't the coolant be a little more... enthusiastic about cooling? Actually grabbing onto some of that heat and doing something with it? You're talking about a temp reading on a tiny amount of silicon which has very little mass and thermal conductivity which interfaces with a hunk of metal that has many times more mass. The core is going to hold on to a lot of heat under load, and those numbers are useless unless we have other chips to compare against.
|
# ? Jun 19, 2017 23:11 |
|
Looking at the various benchmarks I believe that the reason why Intel is clocking SL-X so high is the smaller L3 cache. It seems to impact gaming performance negatively, so much so that Broadwell-E could be faster than SL-X in gaming and at comparable frequencies. That'd also explain why they seem to be happy with the ridiculous thermals. (7900X using 70W more than 6950X at same rated TDP)
|
# ? Jun 19, 2017 23:30 |
|
|
# ? May 24, 2024 19:41 |
|
eames posted:Looking at the various benchmarks I believe that the reason why Intel is clocking SL-X so high is the smaller L3 cache. It seems to impact gaming performance negatively, so much so that Broadwell-E could be faster than SL-X in gaming and at comparable frequencies. Aaaaand my 5820k OCed to 4.5 is suddenly looking a lot more competitive for sane-man uses.
|
# ? Jun 20, 2017 00:12 |