Read more.Quote:
Quartet available April 11 from $169.
Printable View
Read more.Quote:
Quartet available April 11 from $169.
From a marketing perspective wouldn't it be better for AMD to pitch the 1600X as it's answer to the 7700K?
A 6C 12T 1600X for £250 would look & sound a lot better to a potential buyer than a 4C 4T 7700K for £330.
Okay, the 1600X will still lose in single threading performance but those 50% extra cores and threads will still ensure better performance in most multi threaded apps and games or where the 7700K's slightly better IPC & much higher clock speeds will at best only match the 1600X in such tasks.
I've always seen these 6 core parts as the proper rival to the 7700K, not the 1700, as AMD seem to pitch.
From a commercial point of view it would be better for AMD if the 6 core chips could be 4+2, 3+3 or 2+4 parts as it would enable them to use more of the ones that didn't make the 8 core grade during manufacture.
However, because of infinity fabric connecting the CCX's being a bit of a bottleneck I can't see them being able to do so cause the 4+2 chips would perform differently to the 3+3 chips under heavy loads.
I think the 6-core on will tempt me. I'd like more cores than my current i7 4790K but without losing too much in the way of clockspeed for older applications. I don't currently do enough multi-threaded tasks for R7 to be worth the extra.
With regards to marketing, wouldn't it have made far more sense for AMD to go down this route:
R4=4 core (and it's bigger than i3)
R6=6 core (and it's bigger than i5)
R8=8 core (and it's bigger than i7)
Did they want to avoid the unlucky number 4 in China that much?
What have I missed with the most expensive CPU not including a stock cooler?
The 1600X has nearly 50% more heat to dissipate, little point charging the consumer for an inadequate cooler. Having said that my I5 2500K (also 95W) came with a cooler although it only proved useful as a paperweight (and a great way to test the buzzer on my PC when running multi-threaded applications).
While you are here you might as well check out the AMD Ryzen 5 Desktop Processor Sneak Peek video:
https://www.youtube.com/watch?v=yWdhLXl5a5s
A 6 core 12 thread processor at 65 watts is good! Seriously tempted to make a nice build this summer now once it's all settled down. ;)
AT thinks that the R5 1500X has 16MB of cache and the R5 1400 has 8MB of cache:
http://www.anandtech.com/show/11202/...n-5-april-11th
So it seems the R5 will be a dual CCX design.Quote:
We have confirmation from AMD that there are no silly games going to be played with Ryzen 5. The six-core parts will be a strict 3+3 combination, while the four-core parts will use 2+2. This will be true across all CPUs, ensuring a consistent performance throughout.
Wrong thread.
Does that even make sense? I thought it would be cheaper manufacturing wise to make a dedicated 4 core 1 CCX part.
I had just assumed when they said that the design was scale-able that you could have 1 CCX parts but maybe the design means you have to have at least 2!?
Its a real big shame that the fastest 4 core part turbos to 3.7GHz....I know that's pretty much what we were expecting but was hoping for more...
huh, no R3 at all? Either yields are terrible or AMD is scraping the bottom of the barrel and can't afford another set of masks for a single CCX, a chip with half the cores gated off is very unusual
The 1500X is an oddball - it's got a 200MHz XFR boost, given how low the power consumption is for the 8-core chips I'd expect it to act like a 3.9GHz CPU in practice
Or a more optimistic possibility is that the 4 core mask was due later anyhow, and in the meantime they found some easy fixes which allow that those parts to clock higher, consume less power or similar. While not expecting that much from GF's 14nm FF in terms of clocks (the efficiency sweet spot seems to be 3GHz is which is good for 8C/16T and ideal for Naples' 32C/64T or mobile) maybe they found some easy fixes which don't need to wait for Zen+
My expectation would be that the R3 is going to be a single CCX part (I was really hoping that the 4/8 R5s would be).
If so the question on my mind now is: Will a single CCX R3 overclock a bit better than a dual and therefore have a higher IPC than R5/R7...
Is the fastest gaming CPU (in the short term) going to be R3 rather than R5 or is it more likely that the latency issues with dual CCXs are mostly surmountable (in firmware/software) and that will let a 3+3 R5 achieve full potential.
Anyone got a crystal ball handy?
This is all great, but when will we have motherboards? I pre-ordered an Asrock Taichi and 1700X ages ago, still not here :(
Seems the motherboards are really lacking in quantity at the moment from all the brands.
If you look at the die shot, it is a wide rectangular part. The CCX complex is rectangular, so why didn't they put them one on top of the other to make the overall die more square? To go for a side by side layout I assume they must need to expose more edge. Think about it, for a given area long and thin is all edge, square gives the least perimeter. That perimeter is where the I/O pads go for all the PCIe, memory etc interfaces. That is called pad limited, where the number of pins you have to wire up defines how much area the chip is going to have.
So if it is pad limited, then that area is going to exist whatever, you may as well find something to put in it. That should soon be graphics cores for the 2/4 core parts, otherwise it may as well be cpu cores, leaving it empty is silly.
I had a 1700x and asus x370 pro pre-ordered, but between the funky gaming benchmarks, the long wait for the motherboard and also corsair taking their sweet time giving an ETA on the mounting bracket for a h100i, I ended up switching to a 7700k and maximus IX hero instead, I don't regret it. I still hope AMD do well though, competition is always good.
Also the post mentioned that the cores are symmetrically disabled, does that mean that they can potentially be re-enabled, like some of the old tri cores?