Review: AMD Ryzen 7 1800X processor

1800x in Romania is 566 euro wtf.
1700 is 355 euro.
Il stick with my fx8350 and 4770k till prices go down.
Maybe an 1600x with 6/12 core at 4ghz is better and cheaper.
Anyway im excited these are the new BARTON.:banana::banana::banana::banana:

 

tech Deja Vu

--A ton of hype leading up.
--Promising benchmark leaks appear, fueling the hype/hope.
--Ultimately we get them to market and it's not a game changer.

Overall, Summit Lake is feeling like a typical AMD release.

 

It's really not. EDIT: The high speed link is there sure, intel has one too.

IF there was a bottleneck as you claim it would be there for everything, not just games.

Zen is very early but the potential is clear.
When Hammer was first introduced it had it's problems too, Northwood would beat it. Time is key, it's up against a very mature platform.

 

I think most relevant comparisons must be done on a price basis. If 7700k = $320, then AMD 1700 would be the chip to compare.

 

I think it actually matched the hype in everything but gaming.

 

AMD have troubles in havy multi-threaded games, that scales well over 4 corest like AOS, when 4 cores are trying to get data from the other bank of L3 cache of the other 4 cores. Maybe Hilbert can run some game benchmarks using only 4 cores, to see if I'm right or not

 

Thank you Hilbert for another great review
Although Ryzen and it's eco system still needs some polishing, i'd say it's a good start and AMD did not disappoint.
i hope with fixes and firm ware update gaming experience will improve :nerd:
still a great value product from AMD, and a most welcomed game changer in CPU market :nerd:

 

excellent work boss. prompt & thorough review. i dont care how nerdy im gonna sound saying this: going through all the bench results was a treat. i was riveted. if this is right out of the gate after engineering samples...oh man! im very excited for where zen is headed once the silicon matures.

ender, i appreciate your intellectual curiosity, but dont you think you might be drawing conclusions prematurely? i think (?) we have all seen the slides leaked from the circuits conference that showed the cache pool for each CCX (the cluster of four cores) with its placement clearly implying that the clusters must be bridged in the upper level metals of the chip itself.

so this isnt a surprise, is my point...but having said that, how do you know the latency isnt just OS scheduler errata? or microcode optimization problem? or a full blown BIOS bug?? we dont actually know, from this ~30ns of unexplained memory latency, that theres anything outright wrong with the cache functionality or the memory controller at all.

occams razor would suggest that its more likely to be a symptom of an incredibly immature platform...especially considering the mobo manufacturers have obviously been having problems getting higher bandwidth DDR4 stable right out of the gate.

if youre right in your assumption then unfortunately youre probably correct to say it cant be fixed - yet frankly id be very surprised if this latency issue (& the gaming framerate thing) is still around after early BIOS revisions & perhaps an OS patch...

 

Where is the 8+8 coming from?

Everything I have seen is it's 16MB shared.

 

Did DooM received any pre-patch for AMD Ryzen? In Vulkan mode the R7 1800x behaves very well with that game.

If the game did not received any particular attention for Ryzen, then doubts against the graphics drivers will increase a lot, by +50%.

 

It looks like 1700 overclocks just as poorly/well as 1700X and 1800X. Doesn't look like there's much point in buying anything else that 1700 if you plan to overclock.

Also it looks like the software (bios) needs work. Windows might need some kind of update too.

 

The R7 1700 looks sweet, especially with under fanless cooler. If I had money to spend, I would buy that model.

 

Rumour is Xbox Scorpion is a Zen variant with Vega GPU part too.

 

Oh, man. Are you like your avatar?

Get better soon.

Great review.

 

it is. but as i mentioned, the CCX is a cluster of four cores & the cache pools for those four cores. zen chip layout shows two CCX adjacently joined together with the cache pools bridged somehow.

just one more thing. nobody knows anything about the interconnect. dont you think its more reasonable to postulate that perhaps the problem is coming from the OS scheduler inefficiently (un)balancing the cache pool or even just the SMT because AMD has completely new branch prediction now??

from where im sitting, that makes more sense than immediately faulting the cache itself & saying that the memory controller is basically gimped as well. but if im wrong, ill be happy to admit it.

 

A few of my old grey haired overclocking mates from years gone by have risen from the grave and had chips for a couple of weeks now.

The same thing is being said...4.2 on air/water is about it.

Might be the case with R5 and R3 too.

 

Sorry I missed this and I see where he is coming from now but it's nothing like an mcm.

Interconnect will be interesting for sure, can't wait to see what gets exposed there.

 

I am glad AMD could finally deliver a product that makes customers happy.
Living up to spoiled customers expectations is a huge milestone.

Regarding low performance in gaming, it seems results vary on motherboard that reviewers used.
Gigabyte motherboard reviewers seem to get higher score in gaming. Customers on reddit confirming it so far.

I am pretty sure we not seing whole picture. Once motherboard socket stables and drivers fix issues, Ryzen might impress us again.

 

Those Ryzens look rather nice. I think I'll wait for the second generation of Ryzen CPUs and chipsets before I upgrade, however. Until then, the kinks should be hopefully ironed out.

 

It seems some reviews have l3 cache latency at nearly 50ns. While we could see Hilbert getting around 20ns.