In order to bring you guys a great Zen 1700 and Gigabyte Gaming 5 review....

[Ket]

I want your suggestions on how you would like me to test Zen for its gaming performance at 720p and 1080p. I've already got a nice selection of tests to run that I consider to be unbiased and representative of real life scenarios but I'm always open to suggestions So have your say, I'll then make a poll and the most voted for options will get included

I'm also preparing for a proper serious OC session with Zen once the review is done. I've already got a 0.5mm 10W/m-K graphite based thermal solution at the ready, in case that isn't suitable due to how the heatsinks may be mounted I've also got a 1mm 6W/m-K Boron Nitride filled silicone elastomer on standby - the same stuff NASA used on the space shuttle I also just received today a AM4 mounting bracket from Cooler Master for the Hyper 212+. I'll be testing if such a cooler is really any better than the supplied Wraith Spire cooler (I think we all know what the outcome of that is going to be ) Unfortunately I do not yet have a bracket for the Cooler Master Seidon 120v to test that out, CM are still working on it.

[chew*]

The medium used for interface will not solve the lack of mass or design. Your over thinking it. All companies should be using finned or pinned heatsinks oriented in a fashion that follows simple logic heat rises.

Stick a fan on it and be done with it. Its proven to work. 1700 does not ramp pwm heat up as bad anyway.

Since most reviewers show ryzen trailing a tad behind in games at lower res showing how you can tune ryzen to scale would be usefull to the audience. For that your going to need a different board and push ref clock.

[vario]

I want your suggestions on how you would like me to test Zen for its gaming performance at 720p and 1080p. I've already got a nice selection of tests to run that I consider to be unbiased and representative of real life scenarios but I'm always open to suggestions So have your say, I'll then make a poll and the most voted for options will get included

I'm also preparing for a proper serious OC session with Zen once the review is done. I've already got a 0.5mm 10W/m-K graphite based thermal solution at the ready, in case that isn't suitable due to how the heatsinks may be mounted I've also got a 1mm 6W/m-K Boron Nitride filled silicone elastomer on standby - the same stuff NASA used on the space shuttle I also just received today a AM4 mounting bracket from Cooler Master for the Hyper 212+. I'll be testing if such a cooler is really any better than the supplied Wraith Spire cooler (I think we all know what the outcome of that is going to be ) Unfortunately I do not yet have a bracket for the Cooler Master Seidon 120v to test that out, CM are still working on it.

Do the tests with some quality ram preferably at 3200 cas 14.Set windows to high performance mode or just do the tests at a fixed 4ghz ,preferably do the SMT on vs off case .
If possible show us how ram scales performance from 2133 to 3200 (3 points should be fine

[chew*]

Smt off is not happening until i get bios for 1700 sorted vario.

Seems its somehow related to xfr or the lack of. X chips do smt off fine.

[Ket]

The medium used for interface will not solve the lack of mass or design. Your over thinking it. All companies should be using finned or pinned heatsinks oriented in a fashion that follows simple logic heat rises.

Stick a fan on it and be done with it. Its proven to work. 1700 does not ramp pwm heat up as bad anyway.

Since most reviewers show ryzen trailing a tad behind in games at lower res showing how you can tune ryzen to scale would be usefull to the audience. For that your going to need a different board and push ref clock.

Heat is drawn in the direction it is "pulled", without a heatsink, heat rises in the relative "up" direction, a inefficient way but how nature does it as warm air mixes with cool. Heatsinks manipulate that process allowing heat to be drawn away much faster. How quickly that heat is drawn away depends on the quality of the heatsink, the materials its made from, and how efficient heat can be transferred to the heatsink and away from the components. What I've tried to explain here is known as Zeroths law of thermodynamics, this law helps define the notion of temperature. Its not as simple as just sticking a fan over something

[chew*]

A chunk of aluminum versus finned aluminum is a no brainer.

Vertical fins vs horizontal in a case mounted orientation is a no brainer.

An overworked 6 phase and an under worked 4 phase is a no brainer. One is scorching hot one is not.

[Ket]

Naturally, but as so few manufacturers actually use a proper heatsink without making a whole new one it becomes even more important to make what you have as efficient as possible without the need to add fans which just adds additional noise and dust, a fan is the fastest way to optimise a heat source.. but its also quite crude. If manufacturers even just started using thicker bases for their heatsinks and better than "middling grade" thermal pads on critical components (which typically are no better than about 3.2W/m-K) that would help a decent amount. Typically a good thermal pad alone even with a rubbish heatsink will often yield good improvements in load situations. With things like VRMs depending how good / bad the stock thermal pad / TIM is, its not unusual to see improvements of around 10c or so even with a poor heatsink. Make a heatsink yourself, the difference can be night and day. I made a heatsink for a Sapphire Vapor-X card not long ago, I removed the terrible stock VRM heatsink and thick-as-a-wall thermal pad with something I made myself using a Phobya XT 7W/m-K thermal pad, VRM temps under load dropped from 119c to 88c, and my heatsink was significantly smaller than the POS Sapphire stuck on the card.

Asrock used to be very good with their heatsinks, solid, robust, thick things, but I have to say the heatsinks on their AM4 boards look flaky to me from what I've seen, but every manufacturer does with their AM4 boards its not just Asrock. I may go in to more specifics about how people can thermally optimise when I start messing around with things fully it depends how much time I have.

[chew*]

They all were decent on 790,890, and Am3+.

Somewhere along the line they lost there way.

I have used quality thermal pads but....the limitation of there quality is what is dissipating the heat. 290x ref cooler pads are pointless.

Aquacomputer waterblock with backplate not pointless.

In this case not enough mass not enough effeciency in design and no heatpipe.

Waste of money.

[Ket]

I agree, I think manufacturers losing their way is them whats known as "costing down", basically how cheap can they go while still being able to call it a "heatsink". The good news is if you buy a quality thermal pad you won't ever have to really worry about replacing it, unlike some TIMs where their properties break down very quickly *cough* MX5 *cough*.

[chew*]

I think in this case its just the cheapest out the door. The heasink has been tooled for a heatpipe. Neither g5 or g7 in pic on site use it nor can they (8 pin in way). Im sure if i go look at there intel boards we will find that heatsink in use with a heatpipe on most likely a board not known to roast pwm.

So basically just repurposed to toss on g5-g7 and kicked out the door as cheaply as possible.

[Ket]

Oh yay..... sounds like I'm going to have "fun" :/ Please tell me Gigabyte are at least fixing the heatsinks to the board properly not using crummy spring loaded push pins...

[chew*]

I checked contact was good. Spring loaded screws.

[Ket]

Not terrible I guess, depends how they designed the heatsink mounting but it sounds like they haven't completely ballsed it up although that shroud I see.. is that seriously a piece of freaking plastic on top? I hope to god its metal.. if not, Gigabyte know that thing will likely melt and/or catch fire, right?

[Ket]

Gaming 5 is in for review, hopefully I can get everything done before the end of the week.

[drmrlordx]

Just my $.02 on heatsink material:

In standard rounded pin orientation - one of the simplest setups you study in heat transfer texts - material doesn't make THAT much of a difference unless you are:

a). restricted by pin length/width
b). restricted by pin density
and/or
c). blessed with lots of airflow

Aluminum vs. Copper, for instance: if you have a pin array of aluminum, you generally have to make them longer given equal pin radius to get the same cooling, though if you want to make shorter copper pins work, you might need more airflow for that to happen. Either way, beyond a certain length, those copper pins will reach near-ambient and do no good versus a shorter-pinned sink or versus a sink with pins of equal length made of aluminum.

Airflow is ultimately your limiting factor. That determines how much heat can be carried away from the sink given a desired delta T between the ambient air temperature and the base of the sink. It also sets your h value. In passive situations things get more complicated . . .

[Ket]

True, copper vs aluminium doesn't make too much difference but it's worrying when manufacturers choose to use aluminium heatsinks then decide to also make them essentially low profile... on Mosfets of all things. Sticking a lump of plastic over those mosfet heatsinks essentially reducing airflow to as good as zero just all adds up to Gigabyte clearly let a blindfolded monkey design their cooling. You also need to consider there are differing grades of quality with some materials that need to be considered as well, such as a low grade copper mixed with perhaps aluminium more than likely isn't going to perform any better than standard grade aluminium alone. I've already tended to as many of the bad thermal decisions Gigabyte made as possible. Short of designing my own heatsinks it won't be getting much better.

[chew*]

True, copper vs aluminium doesn't make too much difference but it's worrying when manufacturers choose to use aluminium heatsinks then decide to also make them essentially low profile... on Mosfets of all things. Sticking a lump of plastic over those mosfet heatsinks essentially reducing airflow to as good as zero just all adds up to Gigabyte clearly let a blindfolded monkey design their cooling. You also need to consider there are differing grades of quality with some materials that need to be considered as well, such as a low grade copper mixed with perhaps aluminium more than likely isn't going to perform any better than standard grade aluminium alone. I've already tended to as many of the bad thermal decisions Gigabyte made as possible. Short of designing my own heatsinks it won't be getting much better.

I don't think it was a monkey necessarily lol. I just think they borrowed an existing heatsink from another product line which probably does not run hot. I am actually going to snoop around. Be back in a few.

The sb or whatever it is does not run hot enough to be a major issue.

[chew*]

Mystery solved.

Tooled for heatpipe - check.

Dimensions and design- check.

Non clue if socket 1151 PWMs run warm. Gaming 270x But I do know that socket 1151 is for 4c/8t max.

[Ket]

Yeah I saw the telltale signs of where a heatpipe would go when I dismantled the cooling, Gigabyte could of at least repurposed a few beefier heatsinks if they didn't have any ready with a heatpipe though. It makes no sense to intentionally design a heatsink thats less efficient than one that follows standard basic thermodynamic rules but that is what Gigabyte has done with the chipset heatsink. The chipset may not get very hot, but why design a bad heatsink for it? Looks and looks alone would be my bet. Thats totally illogical and nobody likes something getting hotter than it needs to, that heat still builds up.

[chew*]

that not all as you use it just feel temp of the 6 phase vs the 4 or install the temp sensors in the heatpipe holes or use an infrared whatever. Sure any of the 3 methods you will notice a difference in temps between the 6 and 4.

[dasa]

some good ideas above and it will be great to see just how much you can improve the ipc in cache\memory sensitive software like games although you may need to wait for bios updates to get the most out of it

a lot of reviews seem to be getting bashed for testing res below 1440p to 4k as they don't believe people should run these res on high end systems
having seen the effects of gpu limited testing leading to many people thinking slow ram vs fast ram makes no difference i have been trying to think of a way to show the flaws in this mindset

perhaps somebody could try ruining at least 2 tests with a line graph showing gpu usage vs fps at different res
the other thing people are saying is the fps are high enough it doesn't mater what cpu performance you have but they don't seem to understand this can vary dramatically from one part of a game to the next even just by turning around to face a different direction you can go from a gpu limit at high fps to a cpu limit with low fps
some people will never be happy but maybe some food for thought

[Ket]

that not all as you use it just feel temp of the 6 phase vs the 4 or install the temp sensors in the heatpipe holes or use an infrared whatever. Sure any of the 3 methods you will notice a difference in temps between the 6 and 4.

The 4 phases actually get hotter than the 6 phases, 36.3c vs 34.5c @ idle, measured with my DT8380 infrared temp gun. Those temps are from my Boron Carbide pads and getting the heatsinks to make proper contact. Stock pads and contact would probably see 5-7c higher idle temps.

some good ideas above and it will be great to see just how much you can improve the ipc in cache\memory sensitive software like games although you may need to wait for bios updates to get the most out of it

a lot of reviews seem to be getting bashed for testing res below 1440p to 4k as they don't believe people should run these res on high end systems
having seen the effects of gpu limited testing leading to many people thinking slow ram vs fast ram makes no difference i have been trying to think of a way to show the flaws in this mindset

perhaps somebody could try ruining at least 2 tests with a line graph showing gpu usage vs fps at different res
the other thing people are saying is the fps are high enough it doesn't mater what cpu performance you have but they don't seem to understand this can vary dramatically from one part of a game to the next even just by turning around to face a different direction you can go from a gpu limit at high fps to a cpu limit with low fps
some people will never be happy but maybe some food for thought

There are a lot of people out there that have very high end systems but game at 1080p simply through personal preference, 1080p results are not only a necessity of any review, but when testing a CPU are even more relevant as those 1080p tests double as a yardstick for CPU capability. They are definitely necessary in order to give a complete picture.