AMD Ryzen 1700 on the Test Bench Under Water

hondaman

The OG Hondaman
Joined
Jul 31, 2004
Messages
71
I'm interested in this. I cancelled my Amazon preorder because Amazon sucks at preordering, come to find out. Now I can re-evaluate if I want 1700 or 1700x
 

JosiahBradley

[H]ard|Gawd
Joined
Mar 19, 2006
Messages
1,791
If this OCs manually just as good as the 1700x whats the real difference? XFR doesn't seem to get us to something crazy like 5Ghz on one core or anything. Looking forward to this one. Too bad/good already got a 1700x on the way.
 

alamox

Gawd
Joined
Jun 6, 2014
Messages
596
there have been alot of talk about Ryzen gaming performance and L3 cache latency, hardware.fr and some posts in anandtech forums suggest that it could be due to win10 scheduler swaping threads outside of Ryzen CCX.
i know that alot of ppl would be interested like me in seeing some tests done on Win7, as an extra bit of information to confirm or rule that out, comparaison tests of few motherboards also would generate alot of interest.
anyway can't wait for your results.
 

KazeoHin

Supreme [H]ardness
Joined
Sep 7, 2011
Messages
8,060
Woo hoo! Let's see what our lord and savior Kyle Bennet can get out of this chip! Make that bitch cry!
 

Nukester

[H]ard|Gawd
Joined
Mar 21, 2016
Messages
1,428
now THIS grabbed my attention. Can't wait Kyle.

I remember back in the day I cooled my AMD chip with a VapoChill. Would love so see some of that love again.
 

sir-gold

Gawd
Joined
Jan 19, 2006
Messages
931
If this OCs manually just as good as the 1700x whats the real difference? XFR doesn't seem to get us to something crazy like 5Ghz on one core or anything. Looking forward to this one. Too bad/good already got a 1700x on the way.
It looks like the amount of XFR overclock (50mhz vs 100mhz), and possibly different binning, are the only differences between the 1700 and 1700x.
The only thing that really matters is whether or not the 1700x is binned higher, since XFR is disabled during a manual overclock anyway.

It does appear to allow manual overclocking on a per core basis though, so maybe it will be possible to get 1-2 cores running at 5ghz at the cost of underclocking/disabling the rest of the cores. (Assuming the motherboard allows it)

I would wait for someone else to try it first, since it seems like a good way to burn out a core. (Maybe Kyle will be willing to try it, once the regular tests are over)
 
Last edited:
Joined
Jan 18, 2008
Messages
1,011
What the hell is that blue "e" and goofy shopping bag pinned your task bar Kyle...? :confused:

Nice temp. Per core OC would be sweet. Too bad I'm not due for an upgrade anytime soon, always liked AMD best! Glad to see some excitement again, long overdue.
 

SonDa5

Supreme [H]ardness
Joined
Aug 20, 2008
Messages
7,403
Nice results with the lap job. Dooes your water block make contact with the entire surface of the IHS?
 

GeForceX

Supreme [H]ardness
Joined
Mar 19, 2003
Messages
4,172
Loving the mirror finish -- well, sort of a mirror finish. As long as it's flat, then it's going to be good.

Curious what the before/after temps were?

Also, would you be re-benching the 1700 with HPET disabled, Windows' power profile set to high performance, updating the BIOS to the latest firmware, overclocking the memory speed, etc. These would make big differences for CPU benchmarks.
 

Stimpy88

[H]ard|Gawd
Joined
Feb 18, 2004
Messages
1,271
Quote Removed by Kyle, did not want incorrect information in this thread.....


Great work Kyle, but it makes me sad that when you buy a heatsink/waterblock that it is simply not made properly, for a heatsink to be flat is surely heatsink 101! Why the hell is this not done in the factory before it gets shipped to the custer?

But my next question would be, is the IHS on the CPU flat?
 
Last edited by a moderator:

geok1ng

2[H]4U
Joined
Oct 28, 2007
Messages
2,129
This is still a hobby for Kyle. Ryzen did not suffered from high temps, yet he invested time and effort for this lapping.
For those interested in mirror finishing, i just found out that the Brasso metal polisher i sued in my country is available in the USA as well.
The power figures of Ryzen overclocks are quite wide, some reported 216w, others 305w. How about a 3 point Vcore- System power draw experiment with the 1700?
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
49,896
Well, pulling the CPU out to check it with another processor, I did get to check my mating. It felt VERY good last night on install. I have to say I am happy with this result.

Lapping Result 1.jpg Lapping Result 2.jpg
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
49,896
But my next question would be, is the IHS on the CPU flat?
I checked both with a straight edge, and yes, the CPU is very flat. That block is OLD and was convex. I am not sure if the CPUs it was made for were not a tad bit concave. Do not recall. Fixed now though.
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
49,896
That sounds right. I remember Koolance and somebody else had convex Intel and AMD coolers back around the time I bought my Q6600.
Thinking the same thing......I knew it was not "good" when I used it for testing, but did not have time to make it right before leaving for San Fran. That is why I did not make a big stink about temperatures in the review. Wanted to get it all fixed first.
 

THUMPer

2[H]4U
Joined
May 6, 2008
Messages
3,242
I think your Asus board is toast. I've seen a few others with bricked boards.
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
49,896
I think your Asus board is toast. I've seen a few others with bricked boards.
Previous problems mentioned were 100% my fault, so I deleted. OS issues. I have been playing with this install trying to get some of the "SMT issues" figured out, and the OS would not take my commands for some reason to undo, what I had changed. It showed it had, but I think it was the OS. Loading fresh as I type.
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
49,896
Ah, horsing around with bcdedit.exe, groupsize and groupaware?
Yeah, all that and the platform clock. New OS on now. Got all my cores and threads back.

Kyle, dude. I'm not sure how to mention this, so I'll just plunge in. What's with the lipstick stains on your water bottle? Do you do these videos in drag or something? Hey, not that there's anything wrong with that... ;)
Hehe, well, I think that is my wife's water bottle or her girlfriend's. ;)
 

Soarin

2[H]4U
Joined
Jul 23, 2010
Messages
2,491
Gotta say I am super interested in these results..

also Kyle sounds hot :shame:
 

TheHobbyist

Hugs Hard Johnnies [H]ard
Joined
Apr 8, 2008
Messages
456
With regards to the convex mating surfaces on blocks (and also heatsinks), it was a trick used by manufacturers to get even better temps on the processors of the time. If you look at delidded processors, you can see that they are rectangular along the center line. As you bolt down the block/heatsink, the convex aspect pushes down and bows the heat spreader and effectively reduces the distance between the processor and the heat spreader.

There have been a few hardware review sites that realized the heat spreader of their test bench processor had become concave over time testing cpu coolers. Concave heat spreader + flat cooler = worse result. What a mess.
 

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
49,896
With regards to the convex mating surfaces on blocks (and also heatsinks), it was a trick used by manufacturers to get even better temps on the processors of the time. If you look at delidded processors, you can see that they are rectangular along the center line. As you bolt down the block/heatsink, the convex aspect pushes down and bows the heat spreader and effectively reduces the distance between the processor and the heat spreader.

There have been a few hardware review sites that realized the heat spreader of their test bench processor had become concave over time testing cpu coolers. Concave heat spreader + flat cooler = worse result. What a mess.
Yeah, I have actually mentioned this a lot lately as how these new IHS tops are VERY flat compared to what these used to be. You can lay a perfect straight edge on these new AMD and Intel IHS.
 

c3k

2[H]4U
Joined
Sep 8, 2007
Messages
2,159
We'll need a concavity conclave to conclusively sort this out. The concept will be to conclude with a concord so there'll be no more conundrums.

(I'll be here all week. Try the veal. Don't forget to tip your waitress...)
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,265
Bring out the blades and start checkin ;)
I probably will the next time I take the block off the CPU. It looked so shiny and nice when I took it out of the box, that I didn't even think to check it before installing it. I just cleaned it well and plopped it on there. I had thought lapping was mostly a historical concern.

Also, that being said, all this talk is about current CPU's being flat, my CPU isn't exactly brand new either. I wonder how flat it is...
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,265
That, and taking the plastic film off the waterblock helps. :D
Lol. I definitely did that. But I was working with Indigo Extreme Reflow TIM, (I didn't realize I bought a block that included it, so I figured I might as well try it) which I understand is easy to screw up. I wonder if maybe I messed it up when I installed it.

Next time I mount it, I'm going to use some of my Thermal Grizzly Kryonaut and see if I fare any better. I'm too skittish to go with any of the liquid metal TIM's.
 

SonDa5

Supreme [H]ardness
Joined
Aug 20, 2008
Messages
7,403
Looks like there is alot of room for improvement on Ryzen water block design. This is how I would design the block for IHS on R7 series.

1. Entire IHS needs to make contact with block.
2. Identify hot spots of IHS under load with infared camera and design liquid flow through input and output around identified hot spots.
3. All cooling fin channels inside of block should cover entire surface contact area of IHS.
4. Design block dimensions to clear AM4 Socket for bare die mount for those hard core over clockers that will bare die mount.

Somebody build it.
 

Robergal

Weaksauce
Joined
Dec 9, 2005
Messages
119
Kyle me thinks your gonna need a bigger block. :)

Its funny I have a Athlon 64 and an Athlon 64 x 2 sitting in my drawer.
 
Top