• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FRONTPAGE AMD Ryzen 7 1800X CPU Review

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Oh, and when it comes to competative benchmarking, NO ONE uses the same speed for min/max turbo. Look at leaderboards for virtually any game (not synthetics), the leaders are using some combination of less cores at higher turbos and more cores at lower turbos with different variations of turbo staging depending on load. Less threads more turbo, more threads less turbo.

Only time you use the highest minimum turbo setting you can is when you benchmark 3dsmax and Cinebench and the like. Of course those are 16 thread workloads so lowering the min turbo and increasing the max wouldn't make sense given the fact that you pay more power for every mhz every time you increase mhz.
 
Earthdog: I built a Ryzen rig last weekend. Tried to get an Asus ROG Crosshair VI mb but no luck eventhough I had all the other components including a Ryzen 7 1800X. I drove over an hour to the St. David's PA Microcenter to see what they had if any X370mbs. None were in stock but I picked up an Asus Prime B350 Plus mb for $99. Since I had a $50 gift card, my "out" $$ after including tax and a $20 year replacement warranty was @ $70.

I had a Corsair H110i GT AIO cooler available and it worked since it had the 2 sided AMD mount. I had prepared a custom water cooling rig BUT the EK EVO block and backplate caused p[roblems. I've since ordered the EK EVO AMD mount which uses the mb's own backplate. When I finally snag an Asus ROG Crosshair VI mb I'll setup custom water.

I'm running the 1800x stock. On this lowend mb I could OC to 3950 without a problem. However, you lose the XTR feature and since headroom for OCing is very limited ( I think the highest I saw was slightly over 4.1 stable) keeping it stock is not a problem. Solid chip.
 
It's about the cores man!! Can't you read? Mhz and wattage and gaming and stuff......:attn:
But it's not about the MHz, it's about dem GHz!

Best part: Ryzen (assuming we are talking the 1800x, cause it'd have to be) comes with 2 cores @ 4.1Ghz outta the box. Naw brah, naaaw. That's not now turbo speeds work. It comes with 8 cores at the same clock speed, and depending on the current workload, may or may not turbo to the max turbo speed.

Oblig. INTEL TUUUUUUUUURBO BOOOST! (stolen from the Zen rumors thread...)

 
But it's not about the MHz, it's about dem GHz!

Best part: Ryzen (assuming we are talking the 1800x, cause it'd have to be) comes with 2 cores @ 4.1Ghz outta the box. Naw brah, naaaw. That's not now turbo speeds work. It comes with 8 cores at the same clock speed, and depending on the current workload, may or may not turbo to the max turbo speed.

Oblig. INTEL TUUUUUUUUURBO BOOOST! (stolen from the Zen rumors thread...)

Turbo my LN2 pot on that sheet.

1700 (non X) @ 1000 Mhz OC. That's where it's at.

FX-9590 sold at top clocks. 1800X sold at top clocks. OC = Mhz. Not getting Ghz OC from these chips.
 
not really sure I agree with all this can't hit 4.7 on haswell-e lark.. I run 4.5 at 1.26v.. at 1.35 I can run 4.7 fully stable, I just don't want to run haswell-e above 1.3 for 24/7 use, want it to last and crunch work for a number of years..

maybe I got lucky don't know?

been running at 4.5 for a year and half now and can flick a profile on and be at 4.7 for benchmarking in an instant.
 
In later news in AMD thread land, the Haswell-E hit's 4.7ghz. Traditionally a very fast processor indeed. 16 threads at nearly 1050 US dollars is no joke!

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-5960X-vs-AMD-Ryzen-7-1800X/2580vs3916

AMD's low end 1700

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-5960X-vs-AMD-Ryzen-7-1700/2580vs3917

Pay double the price for a little faster processor. Unfortunately AMD is a little late to the game but will sell a ton more 16 thread processors than Intel will at this point in time.
 
In later news in AMD thread land, the Haswell-E hit's 4.7ghz. Traditionally a very fast processor indeed. 16 threads at nearly 1050 US dollars is no joke!

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-5960X-vs-AMD-Ryzen-7-1800X/2580vs3916

AMD's low end 1700

http://cpu.userbenchmark.com/Compare/Intel-Core-i7-5960X-vs-AMD-Ryzen-7-1700/2580vs3917

Pay double the price for a little faster processor. Unfortunately AMD is a little late to the game but will sell a ton more 16 thread processors than Intel will at this point in time.

agreed. I see so many reviewers comparing it to 7700k, it's just stupid. in my mind I was going to be happy with Ryzen at the price it is, if it even got within 10% of my 5960x at STOCK, as it stands, my 5960x at 4.5 GHz, JUST beats the 1800x in Cinebench at STOCK 3.6 GHz.

at STOCK 3.0 my 5960x is like 1300 odd points...

I'm mighty impressed that AMD has gone beyond the 5960x stock vs stock, and it looks from what I've seen if you put Ryzen at 4.0 or 4.1, it will outpace even my overclocked 5960x in some heavy duty tasks.

this to me is the most important point.

as someone who does a heck of alot of multi tasking, and games too, the fact AMD now can do this as well is truly brilliant for half the price. okay, I wish it would overclock further because then it really would pull the high end Intel's pants down completely.

so it isn't even the case of pay twice as much and get a little more performance any more man. you pay twice as much and in some cases get LESS, in some more, and within 2-5% in both. it is an epic time for guys like me who need to run 200+ processes all day every day of the week. =)
 
not really sure I agree with all this can't hit 4.7 on haswell-e lark.. I run 4.5 at 1.26v.. at 1.35 I can run 4.7 fully stable, I just don't want to run haswell-e above 1.3 for 24/7 use, want it to last and crunch work for a number of years..

maybe I got lucky don't know?

been running at 4.5 for a year and half now and can flick a profile on and be at 4.7 for benchmarking in an instant.

You've been lucky: most Haswell-E hit a wall around 4.6GHz/1.35v.

Been even luckier myself: running 4.8GHz/1.32v for almost a year.
 
You've been lucky: most Haswell-E hit a wall around 4.6GHz/1.35v.

Been even luckier myself: running 4.8GHz/1.32v for almost a year.

I haven't ever took it above 1.35v myself, it's just so expensive, the risk of killing it etc lol.

I'm not being so kind to Ryzen though.

I'm at 4.1 GHz at 1.425v currently, not tried dialling voltage down yet, came down from 1.45 to this so far, seems stable, just running stress tests now.

this is however with memory at totally default so I'll play with that after. I was curious with just multiplier and voltage what I could get from it.

I went up to 1.45v to try 4.2 GHz, it boots into Windows but even Cinebench is immediate BSOD.

I don't know whether to try up to 1.5v, need to do more research to figure how safe it can be.

it has been under 100% load for 10 minutes so far at 1.425 and Ryzen Master is reporting 79 C max under water. not particularly happy with this either for the speed compared to what I'm used to with all my other chips.

I know it won't throttle / is safe til 95 C but I don't really like going above 80 C max for 24/7.

EDIT - I should probably mention the loop it's in is running a 360mm radiator, and a 240mm radiator, with a stock 1070 founders, and all the fans running at a really low speed for minimal noise. maybe could drop this temp a bit if I wanted to put up with more noise.
 
Actually you're wrong. Neither Precision Boost nor Turbo Core 2.0 nor 3.0 will automatically boost all 4 or 8 cores to max turbo speed EVER.

Marketing would like you to believe this, but it's never been this way.

R7 will hit 4100mhz on two cores only, doesn't matter if you soak it in L2N and put 1.9v though it, it will turbo to 4100mhz on two cores and 3.7 on the rest unless you go in and add more settings.

From 7700k Wiki page...

Base Single Core Dual Core Quad Core
7700K 4.2 GHz 4.5 GHz 4.4 GHz 4.4 GHz

What do you think "workload" means? It's talking about how WIDE the workload is?

Here's another chart from Wikipedia...
https://en.wikipedia.org/wiki/List_of_Intel_Core_i7_microprocessors#.22Haswell-E.22_.2822_nm.29

5960x base clock 3.0ghz and turbo rating of 3/3/3/3/3/3/5/5

so it does 3.5ghz on workloads 2 cores wide, and 3.3ghz on anything wider than 2 cores out of the box. A 500mhz and 300mhz boost respectively.

LLMv2bo.png
1800x comes out of the box 3.6, does 500mhz and 100mhz boost on workloads 1-2 and 3-8 cores wide respectfully.

- - - Updated - - -

Oh, pardon, Ryzen it will boost to 4.0 on two cores and 3.7 on the rest, THEN XFR will come in and add another 100mhz if it senses it can.

It will NEVER boost to 4.0 on more than two cores out of the box. You could set it up to do this though.
 
Bleakwise, didn't you just say not so long ago the 1800x is 4.1ghz out of the box on two cores? I'm a bit confused what you are saying. Are you saying that you understand what I meant by depending on workload? Or that the base clock is he processors resting state and that it MAY boost depending on workload?

 
This is what I suspect is going on based on experience with my 4770k and z87-a motherboard that allowed me to go in and set 4 turbo settings for 1-4 cores wide myself.

I suspect that like Haswell and Bulldozer, Ryzen has the ability to borrow wattage from cores that are "parked" (as in non turbo, way below the perf/watt curve shoots vertical, not 100% sleep C state).

If you have 2 cores parked, each one sucking 10w or so, whole package on a nice board, small voltage bump pulling 140w or so, what you end up with is quite a bit of current it can redirect and thus power (volt*curret=watts, we're talking watts) that it can move from the parked cores micro-caps into the active cores.

If the parked cores kick back on, it can't borrow that current anymore, so it has to clock them down to avoid errors. That was the principal behind Haswell overclocking at least.

This is the problem with dialing up the 4 core turbo. It will "park" it's cores at a higher base clock setting, this means that there is less power that active cores can borrow from parked cores. You're also increasing heat, because it has to keep the "parked" cores fed with enough current to keep them moving correctly.

The only thing we have control over is the voltage, and the MHZ of our cores when it's doing workloads 2-8 cores wide. By decreasing mhz we decrease the current sent to those cores, by increasing mhz we increase current to those cores, and by increasing voltage we reduce the amount of current needed to produce the same amount of watts (again, watts=current*voltage).

There is one caveat. By increasing voltage we increase power leakage. It's analogous to water, if you have X current and you widen the input pipe going into a skinny pipe, there is more pressure on the skinny pipe because you're pushing more water through it. More pressure means more leaks, except electrical "pipes" are not "make or bust" type of medium, they always leak and raising voltage raises leakage, also stresses the "tubing" more, "pressure" exists in electrical loops too and just like water loops pressure can enhance the wear and tear on the system. Unless you're over-volting like mad it's probably not a big deal, but it's useful to know for understanding why your CPU heats up at the same MHZ, because as long as it's not leaking like crazy you'll have simliar temps at 3.8ghz @ 1.35v as you would at whatever you're at, AND you'll be able to ramp up a less wide workloads higher because there is more power (less current required to get the same watts) to spare.

- - - Updated - - -

It comes out of the box precision boosting to 4.0 on workloads 2 cores wide, and 3.7 on workloads 8 cores wide, then XFR will apply another boost up to 100mhz after that.

XFR doesn't kick in on parked cores though, so at bes tyou get 4.1ghz on workloads 2 coerse wide, and then 3.8 on workloads 8 cores wide.

- - - Updated - - -

I'm saying that "depending on workload" translates to how wide the workload is at any given time.

- - - Updated - - -

Also because Ryzen is 8 cores, vs Haswell being 4 cores, given a workload 4 cores wide, you have more parked cores you can borrow power from.

- - - Updated - - -

Rather, I should say on 2 cores wide, you have 6 parked cores to "steal" power from instead of 2, or 4 to steal power from on a 4 wide load, even. Given a 4 core wide load on regular Haswell you obviously would have no parked cores to borrow current/power (watts) from.
 
Haswell at least put a cap on the CURRENT it could draw.

So, if 100% current draw couldn't get you to 4.4ghz, increasing voltage was the only way to get it to the wattage needed to do so. 100 current * 1.35v is more watts than 100 current * stock or 1.2 voltage, so it could get there.

Problem is, at least with my chip, 4.3ghz full oad 4 cores gets hot as hell. For 1-2 cores it's entirely manageable.

Luckily Haswell is smart enough to draw less current at higher voltages. I found that at 1.35v the wattage was slightly higher, but very simliar to 1.2v, as long as I stayed at 4.1ghz. So at the end of the day it was still cool even though I turned up voltage, AND I could get higher boost clocks on workloads less than 4 cores wide.
 
Haswell at least put a cap on the CURRENT it could draw.

So, if 100% current draw couldn't get you to 4.4ghz, increasing voltage was the only way to get it to the wattage needed to do so. 100 current * 1.35v is more watts than 100 current * stock or 1.2 voltage, so it could get there.

Problem is, at least with my chip, 4.3ghz full oad 4 cores gets hot as hell. For 1-2 cores it's entirely manageable.

Luckily Haswell is smart enough to draw less current at higher voltages. I found that at 1.35v the wattage was slightly higher, but very simliar to 1.2v, as long as I stayed at 4.1ghz. So at the end of the day it was still cool even though I turned up voltage, AND I could get higher boost clocks on workloads less than 4 cores wide.

What are you even on about?
 
Back