|
Probably be over a year before I can justify the cash to replace this PC. But hey, no early adopter tax is right. That's something to comfort our processor envy.
|
# ? Dec 22, 2016 18:30 |
|
|
# ? May 14, 2024 04:56 |
|
I always think "if its actually good then they should send review samples as early as possible" but it never happens, good or bad.
|
# ? Dec 23, 2016 08:37 |
|
Supposed leak of benchmarks from a reputable French hardware magazine Grain of salt and all, but that's honestly really impressive, with a 3.15Ghz BC, 3.33Ghz Turbo keeping up with the i5-6600 @ 3.5Ghz base and 3.9Ghz Turbo. It looks like 4.2Ghz will be achievable as that's the clockspeed of the quad core prototypes. Also if performance is real expect 150-599$ prices but maybe AMD will be stupid and charge Intel prices for getting this close to Intel performance.
|
# ? Dec 23, 2016 18:48 |
|
Looks like I'll be hanging some more onto this overclocked 5820K of mine.
|
# ? Dec 23, 2016 20:30 |
|
Combat Pretzel posted:Looks like I'll be hanging some more onto this overclocked 5820K of mine. I mean, why exactly would you switch to Zen from a 5820K though?
|
# ? Dec 23, 2016 20:32 |
|
I'm hoping to score a cheap eight core CPU for rendering out stuff, if the IPC was similar enough and the Zen overclocks well. As it looks, the performance boost isn't big enough for the cost of a switch.
|
# ? Dec 23, 2016 20:42 |
|
Combat Pretzel posted:I'm hoping to score a cheap eight core CPU for rendering out stuff, if the IPC was similar enough and the Zen overclocks well. As it looks, the performance boost isn't big enough for the cost of a switch. Would you have switched to a 6900K if it cost $500 or $600 instead of $1k? From what I can tell, a ~$500 6900K is about the best case we can expect out of the top end Zen SKU, and that would be killer if they did!
|
# ? Dec 23, 2016 21:42 |
|
If the chip is reasonably cheaper than intel cpu's, I'll upgrade from my 920. Supposing the chipsets and mobos will be OK this time around.
|
# ? Dec 23, 2016 21:43 |
|
If their per-core performance can match my 4.2GHz 2600k for less than an Intel equivalent, I'll probably go AMD.
|
# ? Dec 23, 2016 21:49 |
|
Combat Pretzel posted:I'm hoping to score a cheap eight core CPU for rendering out stuff, if the IPC was similar enough and the Zen overclocks well. As it looks, the performance boost isn't big enough for the cost of a switch. Performance is Broadwell-E and and almost Skylake power consumption, I'd say it's pretty good, at least for someone like me who will be moving from Haswell quad to a 4/8 or 6/12 Ryzen. I wonder if OEMs are chomping at the bit for Zen CPU's because they can exploit their relationship with AMD more and get better margins.
|
# ? Dec 23, 2016 22:28 |
|
You're all doing the price/performance math, and to be fair, so am I, but based on my inputs, my output is "DEAR loving GOD GET ME OFF THIS loving PHENOM II X4" ALREADY"
|
# ? Dec 23, 2016 22:28 |
|
Is there going to be a decent 4-core option for those of us who don't need 8-cores, but would like to spend less than the cost of a 6600k? Or is it all still up in the air?
|
# ? Dec 23, 2016 22:30 |
|
I'd imagine a 4c8t amd to cost around 200 bux give or take. Make it close to 300 and people will just buy an i7 e: On the other hand, it's entirely possible AMD is going to offload all of their chips onto HPC customers for the first several months, and will just set prices to match intel on desktop and not care at all until much later if they actually sell large numbers.
|
# ? Dec 23, 2016 22:35 |
|
SwissArmyDruid posted:You're all doing the price/performance math, and to be fair, so am I, but based on my inputs, my output is "DEAR loving GOD GET ME OFF THIS loving PHENOM II X4" ALREADY" I'm still on a Phenom II X6, I feel you
|
# ? Dec 23, 2016 22:36 |
|
Twerk from Home posted:Would you have switched to a 6900K if it cost $500 or $600 instead of $1k? From what I can tell, a ~$500 6900K is about the best case we can expect out of the top end Zen SKU, and that would be killer if they did! Also, there's the silicon lottery to consider. I still haven't researched how much drama it is to overclock an 6900K to 4GHz stably. By drama I mean voltage required for a stable overclock. --edit: I think I've misunderstood the question. Either way, I guess it'd be worthwhile to wait for a few more benchmarks. If I read that blurry text correctly, the Zen ran at 3.4GHz turbo, while the 6900K did run at 3.7GHz. If you were scale the benchmark linearly with the clock (which we know doesn't work like that), the Zen would still be 10% short of the Broadwell-E. Combat Pretzel fucked around with this message at 00:17 on Dec 24, 2016 |
# ? Dec 24, 2016 00:12 |
|
Truga posted:I'd imagine a 4c8t amd to cost around 200 bux give or take. Make it close to 300 and people will just buy an i7 Server market is where the money is, if they somehow get to double digit x86 server market share then Dr. Su has saved the company.
|
# ? Dec 24, 2016 00:24 |
|
FaustianQ posted:Supposed leak of benchmarks from a reputable French hardware magazine Had a chance to sit down and think on this. One hopes this is not production silicon, since it isn't 3.4 GHz. One also hopes that Lisa Su isn't going to have to eat crow. Seventeen weekdays left to get it right, boys. Don't gently caress this up. I'm thinking in the end, we'll get 80% of a 6900K, but if they go any lower than 75% of the price of a 6900K before letting Intel move and then undercutting them some more, I think it's bad and leaving money on the table. ....god know how that math works out for Kaby Lake, though. SwissArmyDruid fucked around with this message at 00:57 on Dec 24, 2016 |
# ? Dec 24, 2016 00:53 |
|
SwissArmyDruid posted:....god know how that math works out for Kaby Lake, though. Probably in a fashion summed up "a day late and a dollar short".
|
# ? Dec 24, 2016 01:05 |
|
Clocks are actually 3.15Ghz base and 3.3Ghz Turbo, it's not finalized this was that ES silicon that appeared in the wild like 4 months ago.
|
# ? Dec 24, 2016 03:57 |
|
FaustianQ posted:Clocks are actually 3.15Ghz base and 3.3Ghz Turbo, it's not finalized this was that ES silicon that appeared in the wild like 4 months ago. If the Intel ES chips that crop up on ebay are any indication, production silicon will probably have an extra 10% or so.
|
# ? Dec 24, 2016 04:39 |
|
What's the max amount of RAM on the new AMD platfrom? Or is that still a mystery?
|
# ? Dec 24, 2016 05:03 |
|
I don't think the official RAM limits have been given yet but I'd be surprised if it couldn't handle 4x 16GB DIMMs for desktop usage. Probably won't handle that many DIMMs at a high clockspeed though.
|
# ? Dec 24, 2016 09:44 |
|
FaustianQ posted:Clocks are actually 3.15Ghz base and 3.3Ghz Turbo, it's not finalized this was that ES silicon that appeared in the wild like 4 months ago.
|
# ? Dec 24, 2016 13:43 |
|
Haquer posted:I'm still on a Phenom II X6, I feel you A year ago I finally played musical chairs and pulled the 705e out of the server and moved my 1100T there, and went Intel on my personal desktop for the first time since a 733 mhz coppermine mounted on a Slocket adapter...
|
# ? Dec 24, 2016 17:52 |
|
PC LOAD LETTER posted:I don't think the official RAM limits have been given yet but I'd be surprised if it couldn't handle 4x 16GB DIMMs for desktop usage. What is the reasoning behind this? Does the 6700k have the same limitations?
|
# ? Dec 25, 2016 04:14 |
|
Well the current old FX processors can handle 32GB so it's logical to double that, at least, especially if they want to sell a server version of this thing. Edit: an 8c/16t Intel on an x99 motherboard supports 128GB though FuturePastNow fucked around with this message at 05:14 on Dec 25, 2016 |
# ? Dec 25, 2016 05:05 |
|
ItBurns posted:What is the reasoning behind this? Does the 6700k have the same limitations? It does, mad 64gb ram supported. Which I mean sounds pretty reasonable to me. What's the relationship between max ram/number of ram channels and pins on the CPU again? I recall that might be a limiting factor.
|
# ? Dec 25, 2016 08:16 |
|
ItBurns posted:What is the reasoning behind this? Does the 6700k have the same limitations? Yes the i7 6700K has the same limitations. Supporting large amounts of RAM is something reserved for server/HPC products. Its part product segmentation and part for practical reasons because running all those traces and making the CPU package to support either more DIMMs is expensive and difficult and so is designing and testing a memory controller that will support very high capacity DIMMs too.
|
# ? Dec 25, 2016 23:58 |
|
PC LOAD LETTER posted:Its a desktop oriented product not a server/HPC oriented product + 64GB is still a whole lot of RAM. 32GB is more than enough for most people right now. 16GB is probably the sweet spot in terms of capacity and cost. That will change over time but generally its one of those things that is relatively slow to change. I have 32gb now and it's absolutely crippling. Anyways, I was referring more so to this... PC LOAD LETTER posted:Probably won't handle that many DIMMs at a high clockspeed though. What clockspeed does Sky/Kaby lake support with 4x16gb? What is the reasoning behind the AMD platfrom falling short? ItBurns fucked around with this message at 00:15 on Dec 26, 2016 |
# ? Dec 26, 2016 00:13 |
|
SwissArmyDruid posted:You're all doing the price/performance math, and to be fair, so am I, but based on my inputs, my output is "DEAR loving GOD GET ME OFF THIS loving PHENOM II X4" ALREADY" That was my last AMD cpu, and my favorite. But good god man I see you in GPU threads and gaming threads so promise that no matter what happens get something because the performance difference between my OC'd Phenom 965 BE and a plain jane no OC 4670k was stark. At least in online games iirc. Not that there's a huge difference between today's Intels and that one but it adds up a bit and the phenom isnt getting any faster
|
# ? Dec 26, 2016 00:47 |
|
Combat Pretzel posted:Also, there's the silicon lottery to consider. I still haven't researched how much drama it is to overclock an 6900K to 4GHz stably. By drama I mean voltage required for a stable overclock. My 5820K does 4.13 GHz at stock voltages. PC LOAD LETTER posted:Its a desktop oriented product not a server/HPC oriented product + 64GB is still a whole lot of RAM. 32GB is more than enough for most people right now. 16GB is probably the sweet spot in terms of capacity and cost. That will change over time but generally its one of those things that is relatively slow to change. Even the base 6800k does 128GB. The only segmentation here is that the small-die 2C and 4C chips don't support it - because they don't have a quad channel memory controller. Broadwell doubled it to 32 GB per channel across the board. Paul MaudDib fucked around with this message at 01:51 on Dec 26, 2016 |
# ? Dec 26, 2016 01:31 |
|
Paul MaudDib posted:My 5820K does 4.13 GHz at stock voltages.
|
# ? Dec 26, 2016 02:04 |
|
ItBurns posted:I have 32gb now and it's absolutely crippling. ItBurns posted:What clockspeed does Sky/Kaby lake support with 4x16gb? ItBurns posted:What is the reasoning behind the AMD platfrom falling short? Paul MaudDib posted:Even the base 6800k does 128GB. Paul MaudDib posted:The only segmentation here is that the small-die 2C and 4C chips don't support it - because they don't have a quad channel memory controller. Broadwell doubled it to 32 GB per channel across the board.
|
# ? Dec 26, 2016 02:43 |
|
PC LOAD LETTER posted:Isn't that chip a rebadged Xeon though? Same thing with the 6900K, its just a rebadged Xeon. Pretty much everything else they make still has a 64GB or less limit for desktops. By this metric, so are the 6700K and below, they are market segmentation versions of the Xeon E3s without ECC (except the i3s) and sometimes without hyperthreading. The big-die chips do make up a minority of all the chips Intel sells (in both desktop and server markets), but its not like they're some secret, the 5820K and 6800K are priced right alongside the 6700K. quote:I don't see how this addresses what I was saying. There are practical reasons for production segmentation as I pointed out. This isn't a market segmentation thing, dude. HEDT/Xeon E5 are a totally separate die design, they are much larger. But in general it is the same-ish memory controller, and it can address 32 GB on each channel. The big chips have more channels because they have more die space and more justification since they have 2.5x more cores. Paul MaudDib fucked around with this message at 03:28 on Dec 26, 2016 |
# ? Dec 26, 2016 03:18 |
|
For what it's worth I'm fairly constrained on 16GB
|
# ? Dec 26, 2016 04:03 |
|
Paul MaudDib posted:By this metric, so are the 6700K and below, they are market segmentation versions of the Xeon E3s without ECC (except the i3s) and sometimes without hyperthreading. Paul MaudDib posted:This isn't a market segmentation thing, dude. Paul MaudDib posted:But in general it is the same-ish memory controller, and it can address 32 GB on each channel. The big chips have more channels because they have more die space and more justification since they have 2.5x more cores.
|
# ? Dec 26, 2016 04:28 |
|
Wow you guys use a lot more ram than me. I can barely touch 8 gb and even then I'm pretty sure thats with a memory leak or problem in some way. 16 is nearly wasteful for me but the cost difference was almost nothing for DDR4 at the time 32 gb not enough? VMS?
|
# ? Dec 26, 2016 05:12 |
|
In my case, yes. VMs and CAD. I accept that I have to get Windows 10 eventually, but when I do, it's going to be on MY terms.
|
# ? Dec 26, 2016 05:16 |
|
PC LOAD LETTER posted:Isn't that chip a rebadged Xeon though? Same thing with the 6900K, its just a rebadged Xeon. Pretty much everything else they make still has a 64GB or less limit for desktops. PC LOAD LETTER posted:That doesn't make sense. There are substantial differences between a 6800 and 6700 or below and are physically different dies. The same is not true with a 6800 vs a Xeon. There are consumer counterparts to every single chip in the Xeon lineup. In fact, all consumer chips are explicitly Xeons with some of their features fused off. An i3 is a high-leakage Xeon E3 with 2 of its cores turned off. An i5 is a high-leakage Xeon E3 with its hyperthreading and ECC turned off. An i7 is a high-leakage Xeon E3 with its ECC turned off. There's no debate on this point, consumer chips are the defective runts and datacenters pay big bucks for the pick of the litter. This is how everyone does it, AMD included. FX chips are lovely Opteron rejects just like Intel consumer chips. Yes, there are technical reasons (not every feature is functional on every chip, chips differ in leakage, etc) as well as marketing reasons. But you are totally ignoring the point here, which is that Intel is not cutting anything out from its consumer lineup in terms of max RAM capacity. The max RAM capacity of a Xeon E3 V5 is 64 GB, just like the consumer i3/i5/i7 versions. The max capacity of a Xeon E5 v4 is 128 GB per socket, just like the consumer i7 HEDT chips. The E7 parts, well those don't have a consumer counterpart because there's no consumers who need 3 TB of memory per socket. But for any consumer chip out there - the Xeon has exactly the same memory capacity. That's the physical limit of the memory controller - 32 GB per channel on their scale-out style chips. The E5-based parts have a wider quad-channel memory controller, but it's the exact same controller you get in the HEDT i7 chips (minus the ECC of course). quote:Interesting to note: Kabylake-X is supposed to be a quad core part with 4 memory channels and its a 14nm part just like the i7-6700K. I have no idea how much die space the memory channels pads take up but if die space really was the limiting factor that probably wouldn't be possible short of Intel disabling quite a few of the CPU cores to make a shippable 4 core part which wouldn't make sense unless yields were terrible but I doubt that would be the case. Kabylake-X is supposed to be dual-channel. Look at the cache - it's just a E3 die stuck onto an E5 socket. In fact I would speculate that it may actually take the 6800K's price point - since the 6800K is already priced the same as the 6700K. There's nowhere for a "7750K" to slot in unless the 7800K is getting more expensive. http://www.pcgamesn.com/intel/skylake-x-kaby-lake-x If a quad-channel 4-core part did exist, I think it would be at least $100 more expensive than the LGA1151 parts, because it would be a weirdo specialty part. Paul MaudDib fucked around with this message at 05:39 on Dec 26, 2016 |
# ? Dec 26, 2016 05:26 |
|
|
# ? May 14, 2024 04:56 |
|
Also if anything you should be way more concerned about them fusing off ECC RAM than not putting enough memory capacity on there. Row-hammer attacks are extremely viable right now and open up all kinds of weird attack vectors. In comparison, the number of people who need more than 128 GB of RAM is pretty much zero right now. Even VMs and poo poo are going to run out of cores long before they run out of RAM - and if you are really that edge case then you either buy a dual-socket board (gets you up to 256 GB) or you buy the E7 scale-up processors instead of the E3/E5 scale-out chips (which gets you up to 24 TB). The Zen is going to compete much more with the E5-tier parts than the E3-tier parts. Which is why it supports 128 GB of RAM (like the E5s/HEDTs) in the first place. However, it's only dual-channel not quad. Its memory controller handles more per channel. Paul MaudDib fucked around with this message at 05:41 on Dec 26, 2016 |
# ? Dec 26, 2016 05:29 |