Picked up the Rare AMD 5900OEM for a replacement on my AMD 3600 machine.

[Icon Charlie]

First time making a post here. Today I installed a AMD 5900OEM (not a 5900X) in my infamous AMD 3600 Rig that sort of trolled a few Nubtubbers in 2019 on what you should buy because they got sponsored by the tech industry and yadda, yadda, yadda. The CPU is a Alien Ware system pull.

I'm going to be testing this bad boy for a few more days before giving a final assessment.

But I bought it for $430.00, which is a good deal at the time of purchase to get a 12core/24thread machine.

If I am going to have a concern is going to be temps, as my AMD 3600 was running very cool. Right now do I notice a difference between the two setups in my use so far?

Oh hell yea!!!!

But let me give it a few days so the newness wears off and decide if it is worth the money for the upgrade for my purpose.

I'll get a few picts up too. But my report will be geared for a private tech forum as well another site. I figured I share some of it here.

 

[tabascosauz]

First time making a post here. Today I installed a AMD 5900OEM (not a 5900X) in my infamous AMD 3600 Rig that sort of trolled a few Nubtubbers in 2019 on what you should buy because they got sponsored by the tech industry and yadda, yadda, yadda. The CPU is a Alien Ware system pull.

I'm going to be testing this bad boy for a few more days before giving a final assessment.

But I bought it for $430.00, which is a good deal at the time of purchase to get a 12core/24thread machine.

If I am going to have a concern is going to be temps, as my AMD 3600 was running very cool. Right now do I notice a difference between the two setups in my use so far?

Oh hell yea!!!!

But let me give it a few days so the newness wears off and decide if it is worth the money for the upgrade for my purpose.

I'll get a few picts up too. But my report will be geared for a private tech forum as well another site. I figured I share some of it here.

Nice catch. I'd be very interested to see your results. Normally the non-X OEM Ryzens offer very competitive performance with some PPT tweaking, but I'm curious as to whether 90W PPT really is enough for 12 cores to stretch their legs, or if it's just going to sit close to its 3.0GHz base clock. With any run of the mill tower cooler at 70-75C the 5900X runs about 4.1-4.5GHz all-core.

Hit it with some curve optimizer and open up the PPT a little! I'm running a pretty conservative undervolt and am still able to get all round better-than-stock performance capped at a lower than stock 130W. For a rated "4.7GHz" boost I'm guessing the default global boost freq limit will be 4850MHz, with boost override that will be 5050?

I'd be surprised actually if the 3600 runs cooler than the 5900. Zen 2 was not very good thermally at all. Zen 3 gets rather warm when single-core boosting high esp. under AVX load, but the 5900X is very cool in multi thread even at 142W. Dare I say, about as cool if not cooler than my former 3700X at 90-100W ish.

 

[freeagent]

I wonder if they are like the old i7/Xeon in that the Xeon does the i7 clocks with half a volt less power. Add that extra half a volt and watch them fly..

Edit:

maybe half a volt is a bit excessive.. but you know..

 

[RyzenMaster.sys]

If the 5800 is anything to go off of, performance will be just a couple of percent slower, but it probably is a lower bin than the X skew.

 

[eidairaman1]

I wonder if they are like the old i7/Xeon in that the Xeon does the i7 clocks with half a volt less power. Add that extra half a volt and watch them fly..

Edit:

maybe half a volt is a bit excessive.. but you know..

It would have to be Ryzen pro or a AM4 based TR or EP lol

 

[Icon Charlie]

I'd be surprised actually if the 3600 runs cooler than the 5900. Zen 2 was not very good thermally at all. Zen 3 gets rather warm when single-core boosting high esp. under AVX load, but the 5900X is very cool in multi thread even at 142W. Dare I say, about as cool if not cooler than my former 3700X at 90-100W ish.

This also depends on the quality of the motherboard as it can also affect your temperatures as well. But I am pleasantly happy with the current voltage requirements and fine with the temps. When I bought my 1800x I was in heaven, that was one hell of a CPU for it's time and I was running Win 7 on it.

When I bought the 3600, though it is a great cpu for the budget conscious I noticed a difference when multitasking. It wasn't much but to me it was noticeable.

The 5900OEM so far gives me that feeling once again as everything I am doing is a lot snappier than the 3600. Yes of course its got almost twice the actual horse power than the 3600 in many applications.

What I am curious actually the information of dropping in one CPU for another, with only a bios change and plotting down the results. So there will be no driver changes, etc. Whatever I was doing on the 3600 will be done on the 5900OEM. I've been on it for 12 hours as of this posting and the overall experience has been pretty good.

But everything you purchase, well fits a purpose. I am not expecting the 5900OEM to run rings around the 3600 for gaming. My rig is getting geared towards illustration/video editing, the few games that I do play, and delving into VM/Linux in the near future. I don't need bleeding edge material. Just something I think will be relevant enough for a few more years.

Also I'm not a fan of Win 11 and what Micro$haft is doing.

Anyhow a few days of testing and going through the motions. Logically give some viewpoints between the two and then later start tweaking things a bit. I'm horribly anal about airflow management and keeping things cool on a budget.

And in a geeky sort of way it has given me something to work on

 

[tabascosauz]

What I am curious actually the information of dropping in one CPU for another, with only a bios change and plotting down the results. So there will be no driver changes, etc. Whatever I was doing on the 3600 will be done on the 5900OEM. I've been on it for 12 hours as of this posting and the overall experience has been pretty good.

You should still probably install the latest chipset drivers though. When I swapped in my 5900X for my 3700X, I saw a noticeable improvement in scores across the board compared to running off the old firmware. Especially if you don't usually keep up to date. Also not sure which BIOS you're on, make sure you're on at least AGESA 1200 or later for good Zen 3 performance.

TI am not expecting the 5900OEM to run rings around the 3600 for gaming. My rig is getting geared towards illustration/video editing, the few games that I do play, and delving into VM/Linux in the near future. I don't need bleeding edge material. Just something I think will be relevant enough for a few more years.

Is it so unreasonable to expect though? It is Zen 3 after all. It is the cream of the crop for gaming and single threaded stuff that isn't SuperPI.

This also depends on the quality of the motherboard as it can also affect your temperatures as well. But I am pleasantly with the current voltage requirements and fine with the temps.

Mobo and power reporting deviation affects multi core performance but the ST boosting on Zen 3 is pretty consistent across different boards I find, barring glaring firmware issues with the board.

That all-core stock Vcore definitely helps temps. I've yet to see any 12-core Zen 3 pulling more than 1.225V bone stock, but Zen 2 allowed north of 1.3V while stock.

 

[Icon Charlie]

Okay... Lets start off. My opinions now after putting it through several days of what I would normally do plus some tests. I post some of my results now. My premise is... "what kind of performance I would have if I did the bare minimum in updating. Just replacing the CPU and a bios update. The reason for this is I know in my career that many of people have just done that... Plug and play without updates.

First of all the 5900OEM is the real deal. What I used as paste was the Artic MX5. My CPU cooler is a modified Hyper T2 with a dual fan setup. The fans are the Artic 92mm version and is rated @135TDP with the standard 1 fan setup. This setup worked great on my 1800X as well as my 3600 and it should work well with the 5900OEM. I really like the engineering of having a 2 tube setup acting like a 4 copper tube setup.

With the Dual fan setup and using Artic 92mm fans it is essentially has the same or slightly better performance than a 1 - 120mm fan setup. Why do this? To be as compact as possible as well as the cooler doing double duty. The fans are pointing top to bottom instead of left to right as normally seen. This helps pull air into the case from the top and pushes the air downwards into the back of the video card which gets deflected/pulled into the 140mm fans on the side panel.

My computer case is a Corsair 500r with a 400r side panel. Again a hybrid for my rig. My cases as well as all that I build are made for function over style. I only run 2 - 140 mm Artic case fans on the side panel and the Corsair 120mm Case fan on the top back end and that's it. It's nothing fancy. It just works and works well.


IMHO... The 5900OEM is incredibly snappy and there was some good overall increases of frame rates while playing Overwatch. My go to game. I would play for 2 hours take a break and play for another 2 hours. Total time on the computer averaged 10 hours per day for this test.

On day 3, I decided to update all of my drivers and bios fan set up to see if there was any additional increase in performance.

Here are some of the results.

I took a screen shot and the best of the 3 days in the morning at 8AM@24c and at 5PM when the room reached 30c. No driver updates with these pictures. To be honest all of the data points were only off a few points here and there between each other. Nothing really showing out and going... "hmmmm". So to keep the pictures down I'll choose what I think was the best overall from the tests.

Morning 8AM @24c


Around 5PM when the room hit's 30c on the average. The room where the computer is in the south wall sees the sun. Great during the winter. Not so great in the summer when the room temps can reach to 44-46c. Yes its a hot room... with the fans on....

5PM @30


As you can see the Temp Numbers are pretty good just plopping in the CPU without any driver fixes. Pretty much at the end of the 3rd day I decided to update all of my drivers and play a few more games while testing my wattage from the wall.

10PM @26c


With the updated Drivers and managing my fans IMHO made a difference in in cooling which meant better performance all around. I am extremely happy with the data that I gathered about this CPU and my major worries about overheating has been put to rest.

The hottest the CPU reached after continual for the AMD 5900OEM use was at 79c @30c for the testing was great as the highest I ever reached with the AMD 3600 was 78c @30c.

I have a August 20th snap shot of the AMD 3600 @ 12:54 AM for a comparison.


Twice the core and thread count with pretty much same temp levels. With the updated drivers my MOS and my Chipset is overall the same or of less temps than my AMD 3600 setup. The temps on the CPU were overall higher (average 3 to 5c range) than the AMD 3600 but again still far below the safety levels of running the CPU and plenty of room to make some more modifications later down the road.

More data on the way but for now I need to get some sleep

 

[tabascosauz]

Okay... Lets start off. My opinions now after putting it through several days of what I would normally do plus some tests. I post some of my results now. My premise is... "what kind of performance I would have if I did the bare minimum in updating. Just replacing the CPU and a bios update. The reason for this is I know in my career that many of people have just done that... Plug and play without updates.

First of all the 5900OEM is the real deal. What I used as paste was the Artic MX5. My CPU cooler is a modified Hyper T2 with a dual fan setup. The fans are the Artic 92mm version and is rated @135TDP with the standard 1 fan setup. This setup worked great on my 1800X as well as my 3600 and it should work well with the 5900OEM. I really like the engineering of having a 2 tube setup acting like a 4 copper tube setup. View attachment 217117

With the Dual fan setup and using Artic 92mm fans it is essentially has the same or slightly better performance than a 1 - 120mm fan setup. Why do this? To be as compact as possible as well as the cooler doing double duty. The fans are pointing top to bottom instead of left to right as normally seen. This helps pull air into the case from the top and pushes the air downwards into the back of the video card which gets deflected/pulled into the 140mm fans on the side panel.

My computer case is a Corsair 500r with a 400r side panel. Again a hybrid for my rig. My cases as well as all that I build are made for function over style. I only run 2 - 140 mm Artic case fans on the side panel and the Corsair 120mm Case fan on the top back end and that's it. It's nothing fancy. It just works and works well.


IMHO... The 5900OEM is incredibly snappy and there was some good overall increases of frame rates while playing Overwatch. My go to game. I would play for 2 hours take a break and play for another 2 hours. Total time on the computer averaged 10 hours per day for this test.

On day 3, I decided to update all of my drivers and bios fan set up to see if there was any additional increase in performance.

Here are some of the results.

I took a screen shot and the best of the 3 days in the morning at 8AM@24c and at 5PM when the room reached 30c. No driver updates with these pictures. To be honest all of the data points were only off a few points here and there between each other. Nothing really showing out and going... "hmmmm". So to keep the pictures down I'll choose what I think was the best overall from the tests.

Morning 8AM @24c
View attachment 217113

Around 5PM when the room hit's 30c on the average. The room where the computer is in the south wall sees the sun. Great during the winter. Not so great in the summer when the room temps can reach to 44-46c. Yes its a hot room... with the fans on....

5PM @30
View attachment 217114

As you can see the Temp Numbers are pretty good just plopping in the CPU without any driver fixes. Pretty much at the end of the 3rd day I decided to update all of my drivers and play a few more games while testing my wattage from the wall.

10PM @26c
View attachment 217115

With the updated Drivers and managing my fans IMHO made a difference in in cooling which meant better performance all around. I am extremely happy with the data that I gathered about this CPU and my major worries about overheating has been put to rest.

The hottest the CPU reached after continual for the AMD 5900OEM use was at 79c @30c for the testing was great as the highest I ever reached with the AMD 3600 was 78c @30c.

I have a August 20th snap shot of the AMD 3600 @ 12:54 AM for a comparison.
View attachment 217116

Twice the core and thread count with pretty much same temp levels. With the updated drivers my MOS and my Chipset is overall the same or of less temps than my AMD 3600 setup. The temps on the CPU were overall higher (average 3 to 5c range) than the AMD 3600 but again still far below the safety levels of running the CPU and plenty of room to make some more modifications later down the road.

More data on the way but for now I need to get some sleep

Seems reasonable enough at those ambients.

Since you've already put the effort in, may as well use HWInfo for monitoring going forward, it's the basic standard for 3000 and 5000 CPUs. Much more detailed reporting

For example seeing "80C" peak here in HWMonitor doesn't mean a whole lot on its own. Is it just one of the preferred cores or is it most of the package near 80? Is it constantly at 80, or an occasional gaming spike with averages in the 70s (common for Zen 3)? Since temps on Zen 3 are so variable, does it hit similar temps repeatedly day after day? Etc etc

Not to mention that it would be valuable to get an idea of how the contact is on that relatively rough direct touch cooler. Tctl/Tdie (CPU package) acts as a hotspot sensor, so the "80C" might not be coming from the places you expect.

 

[Mussels]

Okay... Lets start off. My opinions now after putting it through several days of what I would normally do plus some tests. I post some of my results now. My premise is... "what kind of performance I would have if I did the bare minimum in updating. Just replacing the CPU and a bios update. The reason for this is I know in my career that many of people have just done that... Plug and play without updates.

First of all the 5900OEM is the real deal. What I used as paste was the Artic MX5. My CPU cooler is a modified Hyper T2 with a dual fan setup. The fans are the Artic 92mm version and is rated @135TDP with the standard 1 fan setup. This setup worked great on my 1800X as well as my 3600 and it should work well with the 5900OEM. I really like the engineering of having a 2 tube setup acting like a 4 copper tube setup. View attachment 217117

With the Dual fan setup and using Artic 92mm fans it is essentially has the same or slightly better performance than a 1 - 120mm fan setup. Why do this? To be as compact as possible as well as the cooler doing double duty. The fans are pointing top to bottom instead of left to right as normally seen. This helps pull air into the case from the top and pushes the air downwards into the back of the video card which gets deflected/pulled into the 140mm fans on the side panel.

My computer case is a Corsair 500r with a 400r side panel. Again a hybrid for my rig. My cases as well as all that I build are made for function over style. I only run 2 - 140 mm Artic case fans on the side panel and the Corsair 120mm Case fan on the top back end and that's it. It's nothing fancy. It just works and works well.


IMHO... The 5900OEM is incredibly snappy and there was some good overall increases of frame rates while playing Overwatch. My go to game. I would play for 2 hours take a break and play for another 2 hours. Total time on the computer averaged 10 hours per day for this test.

On day 3, I decided to update all of my drivers and bios fan set up to see if there was any additional increase in performance.

Here are some of the results.

I took a screen shot and the best of the 3 days in the morning at 8AM@24c and at 5PM when the room reached 30c. No driver updates with these pictures. To be honest all of the data points were only off a few points here and there between each other. Nothing really showing out and going... "hmmmm". So to keep the pictures down I'll choose what I think was the best overall from the tests.

Morning 8AM @24c
View attachment 217113

Around 5PM when the room hit's 30c on the average. The room where the computer is in the south wall sees the sun. Great during the winter. Not so great in the summer when the room temps can reach to 44-46c. Yes its a hot room... with the fans on....

5PM @30
View attachment 217114

As you can see the Temp Numbers are pretty good just plopping in the CPU without any driver fixes. Pretty much at the end of the 3rd day I decided to update all of my drivers and play a few more games while testing my wattage from the wall.

10PM @26c
View attachment 217115

With the updated Drivers and managing my fans IMHO made a difference in in cooling which meant better performance all around. I am extremely happy with the data that I gathered about this CPU and my major worries about overheating has been put to rest.

The hottest the CPU reached after continual for the AMD 5900OEM use was at 79c @30c for the testing was great as the highest I ever reached with the AMD 3600 was 78c @30c.

I have a August 20th snap shot of the AMD 3600 @ 12:54 AM for a comparison.
View attachment 217116

Twice the core and thread count with pretty much same temp levels. With the updated drivers my MOS and my Chipset is overall the same or of less temps than my AMD 3600 setup. The temps on the CPU were overall higher (average 3 to 5c range) than the AMD 3600 but again still far below the safety levels of running the CPU and plenty of room to make some more modifications later down the road.

More data on the way but for now I need to get some sleep

HWmonitor gives bogus results on Ryzen, use HWinfo64

That's a very weak cooler for Zen 3, anything higher wattage it'd choke easily... which shows how power efficient that 5900 OEM is

 

[pantherx12]

I've got one of these. Clocks nicely.

I set the max ppt to 125w to get 4ghz all core boost and call it a day.

 

[QuietBob]

Around 5PM when the room hit's 30c on the average. The room where the computer is in the south wall sees the sun. Great during the winter. Not so great in the summer when the room temps can reach to 44-46c. Yes its a hot room... with the fans on....

Interesting analysis, thanks for sharing! It seems the Hyper T2 can manage the 5900 even with high ambients. But at this room temperature I'd be far more worried about my own health than the CPU's.
Seriously man, that kind of temperature is outright dangerous to your health and may even be life threatening. Someone posted a video here on TPU where a miner narrowly escaped death after living and gaming in a 40c+ room for a few days.
Take as much care of yourself as you do of your hardware my friend

 

[Icon Charlie]

Interesting analysis, thanks for sharing! It seems the Hyper T2 can manage the 5900 even with high ambients. But at this room temperature I'd be far more worried about my own health than the CPU's.
Seriously man, that kind of temperature is outright dangerous to your health and may even be life threatening. Someone posted a video here on TPU where a miner narrowly escaped death after living and gaming in a 40c+ room for a few days.
Take as much care of yourself as you do of your hardware my friend

Thank you for the comment. And yea in the summer time it does reach those temps 100 to 110f+. But this is my man cave and I've been own this house for 37 years. I have several digital and regular thermometers in the room. If it gets too hot to handle I make sure that I'm hydrated and move. There is about a 2c difference every two feet from the floor to the ceiling so I take advantage of that too. I was hit by a car in 2003 and it later in life it caused me to have severe spinal disease. Dry heat and I'm damned Skippy as the pain goes away, but a High Humidity and heat like in the southern areas of the US, I'm done for.

The Hyper T2 is one of those unusual Coolers I stumbled on. I really liked the engineering to this little cooler. As stated before I slapped it on a 1800X and it kept the CPU cool. So I did my research and added another fan to the unit and it has worked well ever since. I have a 212 Evo with a dual fan setup if this cooling set up goes south for some reason. Another reason for the T2 is to reduce the possibility of wear and tear/stress on the motherboard.

29 years ago I was lucky to stumble in on working at a computer store being managed by an "Idiot Savant" and there are a few interesting stories I can say about working with him. But this guy trained me to repair motherboards/equipment at the time. When the average mobo cost you between $800 to $1200 and higher it made the cost of repairing a motherboard a worth whiled adventure.

The components made now are not as good as they were made 29+ years ago when I started repairing equipment. Where companies will skimp is on thickness of ply, the thickness of the copper/gold/everything on the board. The T2 is light enough for me not to worry about weight when you anchor it on to the mother board in question.

Mind you the X570 MSI A-PRO is your lowest end board on the MSI line. And they are a hot board compared to others.

Also another major heat source in this computer is the AMD 5700/5700XT reference style of video card.

When you are pumping out more heat because you are using lower end components you need to know how to affordably keep your computer as cool as possible.

I'm doing this by not using the current styles of computer cases and with the knowledge of "keeping your computer as cool as possible with what you have". Also UPDATE your drivers on your motherboard. It made a real difference in my case.

I'm able to keep this computer with 3 fans. 2-140mm Artic f14's mounted on the side panel and a Corsair 120mm that came with the case on the top back end of the case.

Right now my cooling options are wide open. I can add on more fans and/or change out my CPU cooler.
But right now the data shows me that the 5900OEM was designed to keep overall prices down for OEM company products for maximum profits. They can use cheaper cooling options and/or components when making their computers to sell to the Consumer.

However this wonderful CPU was cool and stable without any additional driver updates which made it run more efficient and in the Hands of a basic Diy who does not want to deal with undervolting/underclocking your 5900XT, well IMHO, if you have the money, this is the CPU to get.

The Cheapest Base New CPU Pricing as of starting the Test was $565 for the 5900X, $319 for the 5800X, $359 for the 5600X. For the 3000 series the Cheapest Base New CPU Pricing as of starting the Test was $165 for the 3600, $285 for the 3700X, $299 for the 3800X, $429 for the 3900X.

I've been looking for this CPU for 4 months at the right price. What price should you be looking at? IHMO if you can pick up a AMD 5900OEM for under $500 plus Tax and shipping then I think it would be worth wild purchase.

Mine cost 484 with tax/shipping.

I've got one of these. Clocks nicely.

I set the max ppt to 125w to get 4ghz all core boost and call it a day.

Oh I can definitely see that happening

Anyhow more data in a couple of days.

 

[HD64G]

No need to make Ryzen CPUs stay at stable high all-core-clocks for anyone apart from benchmarkers. They boost high when needed and for as many threads as needed. Just set the power limit in the UEFI and leave it handle the loads itself with auto voltage adjusting.

 

[QuietBob]

I'm able to keep this computer with 3 fans. 2-140mm Artic f14's mounted on the side panel and a Corsair 120mm that came with the case on the top back end of the case.

I have the same opinion on case cooling. You shouldn't need a bazillion fans to keep the temperatures at bay. A case designed for optimal airflow should only require a single intake and exhaust fan. This is how I run my desktops, all air cooled. Despite some of the components being power hogs, my temps are low even without A/C.

And sorry to hear about your injury. I'm glad that now you have a powerful PC to play with. Keep us posted!

 

[freeagent]

A case designed for optimal airflow should only require a single intake and exhaust fan.

You mean like Deltas? Most people don’t like the sound of 100cfm and up though..

 

[The red spirit]

Interesting analysis, thanks for sharing! It seems the Hyper T2 can manage the 5900 even with high ambients. But at this room temperature I'd be far more worried about my own health than the CPU's.
Seriously man, that kind of temperature is outright dangerous to your health and may even be life threatening. Someone posted a video here on TPU where a miner narrowly escaped death after living and gaming in a 40c+ room for a few days.
Take as much care of yourself as you do of your hardware my friend

He didn't escape unscathed. He had neurological damage and became disabled for life.

 

[Durvelle27]

It's not rare just OEM. You can find tray processors online or at some stores. Should be fun to play with though

 

[freeagent]

These CPU's are a ton of fun to play with. I can only wonder what it can do with all core clocks.. Linpack Xtreme is impossible for me to run at 4600 with air cooling on my 5900X. The heat reminds me of a Q6600 I bought from someone who won it in a contest from Intel. It did 4GHZ with nearly stock voltage, but the heat was insane. I could not control it even with the window open @ -30 with the thick Panaflos at 12v. That is my CPU with all core clocking, almost impossible to control for me. It literally runs away.. I would buy a 5950 but I cant control a 12 core so really much point unless I was after low core boosts.. which would be fantastic.

Anyywhoo.. its still a sick CPU you have. I am curious what can be done with it, especially on the memory side of things. I am a latency whore. Whatever that means..

If someone here has a D15 and can run Linpack Xtreme with their 5900/5950X at a manual all core clock of 4600-4700Mhz.. let me know please

I am all over the place.. I think it might be bedtime lol..

 

[QuietBob]

You mean like Deltas? Most people don’t like the sound of 100cfm and up though..

Hah, I had those in my Socket A days, bought a couple to experiment with. Boy were they loud

In my testing over the years I learned that the volume of air moved is just one factor involved in proper cooling. I mean, sure, two fans will push more air than one. But if you observe significant temperature drops (>5c) just by adding more fans, then the problem probably lies somewhere else. Either the airflow inside the case is obstructed or badly routed (due to poor case design and/or suboptimal component and fan placement), or the hottest components are not properly sinked. Multiple case fans won't do much for temps if the heat isn't being efficiently transferred from the ICs. That's why I'm an advocate of chunky heatsinks, preferably augmented with multiple heatpipes. They let me run the accompanying fans at lower rpm and hence lower noise levels.

Of course I'm talking about your typical gaming rig in a standard case. PCs used primarily for professional workloads, like serious content creation or scientific research, are a different kettle of fish. When the components are being utilized at close to 100% for most of the day, increased airflow will help with the temperatures. But the basic principles still apply. And those workstations are most often operated in office space with industrial grade air conditioning, which makes much more of a difference than case fans on their own.

Linpack Xtreme is impossible for me to run at 4600 with air cooling on my 5900X

That's a really ambitious task, good luck with it! But I wouldn't say it's something a typical user should be concerned with

 

[Icon Charlie]

I and my wife are old as dirt and we are both disabled. While doing this report she had a major health issue which required all of my attention and while in the process I went down too. I fell twice. Now as you young people can get right up after falling flat on your face, I unfortunately can not and was able to avoid a dirt nap in the process.

So here it is. The goal is to see how much differences there is in wattage performance between two CPU's that suppose to run at the same wattage with the same components being used. With and without driver updates. Secondary was to see the gaming performance between the AMD 3600 and the AMD 5900UM.

So the short term answer is Yes there is a noticeable difference in both heat and wattage but it was not as massive one with CPU temperatures between the two.

And yet the 5900OEM... Yea one of the best deals and perhaps one of the best performance vs wattage use I have ever seen. I am really in love with this setup I have now as this is the basis of how I make all of my builds.

The best bang for the buck.

As said before this is one my computer and this is data relating to my computer, but I stated that I would share my information.


When a open view screen on the 32 inch monitor.
3600 w/o drivers 99watts
3600 w drivers 99watts
5900 w/o drivers 100watts
5900 w drivers 100watts

When using Home media cinema.
3600 w/o drivers 132watts
3600 w drivers 132watts
5900 w/o drivers 133watts
5900 w drivers 133watts

When opening up Brave Browser or Open office.
3600 w/o drivers 101watts
3600 w drivers 101watts
5900 w/o drivers 101watts
5900 w drivers 101watts

When opening up Youtube.
3600 w/o drivers 134watts
3600 w drivers 134watts
5900 w/o drivers 134watts
5900 w drivers 134watts

When opening up Home media and Brave Browser ( I do my reading this way with music).
3600 w/o drivers 134watts
3600 w drivers 134watts
5900 w/o drivers 134watts
5900 w drivers 134watts

When opening up Home media and Youtube (just because).
3600 w/o drivers 144watts
3600 w drivers 144watts
5900 w/o drivers 142watts
5900 w drivers 142watts

When opening up a video downloader and Youtube Down loaded a song 3.48 seconds 480p 5.2mb.
3600 w/o drivers 153watts 3.2 seconds
3600 w drivers 152watts 3.2 seconds
5900 w/o drivers 150watts 2.4 seconds
5900 w drivers 149watts 2 seconds.

When opening up a video downloader and Youtube Down loaded a video song 2.51 seconds 480p 12.6mb.
3600 w/o drivers 138watts 4.2 seconds
3600 w drivers 136watts 3.9 seconds
5900 w/o drivers 136watts 3.6 seconds
5900 w drivers 132watts 3.2 seconds


Video games. Overwatch. Pixeo PX329 32 inch monitor @165mz @ High Mode level of viewing. Windowed down to 27inch for increase frame rate.

NOTE: 1. I saw little change in wattage when I windowed my game to 27inches but I did get higher frame rates. Video temperatures were close enough to call it the same. Highest video Temperatures were @ 74c at 34c room temperature. Highest CPU temperatures was @ 82c at 34c room temperature for both CPU's.

NOTE: 2. My Visiontek 5700 is undervolted Frequency is @1611. Voltage is @ 900. Power consumption is around 130watts. With fan changes my temperatures @ max was at 74c average after 3 hours of continual gaming. Default settings will raise the temperature @ max 91c with only a 3-10fps increase on the 27inch windowed setting.

NOTE: 3 All information is based on my computer and my setup. Every computer is set up differently so what works for me might not work for you. I'm just throwing this up in the forum because I said I would do it.

NOTE:4 This was extremely difficult to do and the result is 60 man hours of my own time working around illnesses of myself and my wife. This data is my own for personal testing. Particularly heating and wattage usage. Average on wattage

3600 w/o drivers 329watts
27inch =160 to 180fps
32inch = 135 to 165fps

3600 w drivers 313watts
27inch =165 to 185fps
32inch = 140 to 180fps

5900 w/o drivers 309watts
27inch =180 to 195fps
32inch = 160 to 185fps

5900 w drivers 292watts
27inch =185 to 220fps
32inch = 175 to 205fps

Playing Overwatch on Epic mode was virtually ½ (and less) of the frames per minute on the information provided above.


The results from all of this...

I was pleasantly surprised on how well overall the 3600 with drivers performed on my computer and it just goes to show you that you really do not need up upgrade your computer for gaming as well as anything else, generally speaking that is. It's something I have stated before on this site. If you got a computer from 2015/2016 and upwards I don't think you will need anything soon.

Where things change is that when you need the sheer raw power that the 5900OEM delivers. The Cine Bench on my computer (R20.060) for the AMD3600 w/update is 3455points. For the AMD 5900OEM w/update on my rig is at 6225points. The Cine Bench point for an AMD 1950 Thread Ripper is 6670.

I have a habit of opening up multiple windows screens. 5, 10, 20 or more windows at a time then listening to music, while leaving other applications open. I visibly see a difference between the 3600 and the 5900OEM. Everything is snappy. From opening the internet to multiple windows, to the things I do for work. The 5900OEM is just wonderful.

The most important thing for me though is wattage and temperatures. From my testing on my computer , yes there was a noticeable drop in wattage (with the kilometer to the wall) using the 5900OEM while playing video games. The bonus is that I got more frame rates as well.

The other thing I want to mentioned is “How” I purchase my PSU and with what rating. I want to keep my PSU as cool and not work as hard as possible. This formula is what works for me.

PSU wattage/2, then times the percentage of PSU rating. So my PSU is Rated at 850 watts and has the 80+ gold rating (or 90% of the wattage @ 50% load). So 850/2 = 425 watts times .90 = 382 watts.

This is the number I use when building my rig as it keeps the computer, cool and IMHO prolongs the life of components in question as PSU is not working that hard.

For myself this was an excellent upgrade that keeps my computer relevant for years to come.

ADDED: 10 days ago I just purchased 64GB OLOY DDR4 PC 4000 with timings of CL 15-16-16-36 @ 1.5 volts. It's got RGB lighting which personally I don't like. I down clocked to run at PC 3600 with the timings of CL 14-14-14-34 @1.45 volts. The timings are the exact same timings that are on my G.Skill Flare X but running at 400mhz faster

The reason I purchased the OLOY brand is that they did have some decent reviews on the company. Secondly they are made from Samsung B die chips and they were on sale for $229.99 for 32 (2x16GB sticks) gigs. Their PC 32 (2x16GB sticks) gigs with the 14-14-14-34 @1.45 volts sticks are 239.99. So not only a savings of 10 bucks I have an nice upgrade ability for my rig in the future.

Anyhow the reason for the 64 gigs is that I am going more and more into video/music editing. Secondly I am going to turn this rig's main OS running Zorin Linux. I like the Windows 7 feel to it. With Zorin installed later as my primary OS I'll be buying a 4th gen NVME stick and place it in it, while having my regular SATA SSD/HDDs used as for storage.

BUT because I have the same timings with the exception of the PC speeds I wondered if there was any significant increases in frame rates while playing Overwatch.

5900 w drivers w 64 GB of ram 292watts
27inch =185 to 245fps
32inch = 175 to 205fps

The only significant increase was top end frame rate at my windowed 27 inch gaming in Overwatch. There was virtually no changes in wattage nor in running the video game in 32 in wide screen. I am also aware that there will be variations of performance due to the quality of monitor as I think running on 1080p (instead of 1440p) with a real gaming monitor would get better results. However I am pleased regardless.

So here is my results. Again my apologies on this but when real life kicks you in the balls, you go and kick theirs back.

I chose to rest up and then give my results later.

 

[GreiverBlade]

I and my wife are old as dirt and we are both disabled. While doing this report she had a major health issue which required all of my attention and while in the process I went down too. I fell twice. Now as you young people can get right up after falling flat on your face, I unfortunately can not and was able to avoid a dirt nap in the process.

all the best to you and your wife and thanks for that thread, really interesting and informative (as i have a 3600 in my rig atm hehe)


remember, "when life give you lemons ... you make lemona.... YOU SQUEZZE THE LEMONS IN HER EYES~!"

 

[Mussels]

@Icon Charlie You'll actually find TPU has quite a few regulars with health issues, young and old. You'll find a lot of tolerance and sympathy for that sort of thing here.

Although... What does "windowed my game to 27 inch" mean? You'd have to specify a resolution for that to help us
(Also lower resolution can increase power draw, if you are GPU limited at higher res)


Also i really want a 5900OEM, damn things are harder to find than unicorns

 

[Icon Charlie]

Although... What does "windowed my game to 27 inch" mean? You'd have to specify a resolution for that to help us
(Also lower resolution can increase power draw, if you are GPU limited at higher res)

I windowed the playing area of the game as if I was playing it on a average 27 inch monitor while trying to keep the same aspect ratio. I took my drafting ruler and measured accordingly. The resolutions on all of this is @ 1440P, because that is still the sweet spot for just about everything.

I took all of my measurements at specific times as listed. It was during a heat spell where I lived at the time of almost all of this.

The main exception is when I added the OLOY ram and ran my tests at the average temps of 27c.

Now as for finding a 5900OEM, you could if you had the money to buy a 5900X and down volt/clock the beast and get similar results. If I heard correctly the 5900OEM were cherry picked as well.

I also believe that the 5800OEM will be another great CPU to have, that is if you can find it.

I'm a firm believer in undervolting recently because of the overall quality/performance the current manufactures are stating. It's like the AMD 5700/5700XT. Those are effing hot video cards to get their performance numbers. I reduced the frenquency by 10% and messed with the voltage output and got good results with only a loss of a few frames per second. This is an excellent trade off if you want to save money on the long run.

Also thank you for your kind compliments.

all the best to you and your wife and thanks for that thread, really interesting and informative (as i have a 3600 in my rig atm hehe)


remember, "when life give you lemons ... you make lemona.... YOU SQUEZZE THE LEMONS IN HER EYES~!"

Again thank you for your kind compliment. But yes I think the 3600 is one of the best overall bang for your buck CPU's out there. As stated I emphasize heat and wattage issues that I believe are just as important as performance.

And because of the continued increases of components and IMHO the reduction of quality of those said components makes this especially important. I strongly suggest that you people get a kill-o-meter and take watt samples and see what your rig is actually using.

Let me throw up some pictures on a few things.

The first picture is what I used to build a better computer than what LTT promoted during the 2019 holidays. And did it for less money with overall better performance and upgradeability. Because of thinking of the future I was able to make that nice upgrade 2 years later when I found the right CPU and later the increase of RAM for my rig.



The next picture will show you that I am using 2 140mm fans on the side, 120mm fan in the back, and my Hyper T2 with dual 92mm fans pulling air from the top of the case downward. I took the grid off of the top of 500r case to shoot this picture. I do not run it open like this.



Here is the reason why I like the Hyper T2 back then. Though it only has 2 loops/4 tubes perse' the contacts are as if it was a 4 pipe version and because it is so light I have no issues of possible MB damage down the road by excessive weight.

You got to remember that a lot of Air Cooled CPU coolers are getting bigger and heavier. This will cause stress to the motherboard IHMO. And considering that I used to repair motherboards when I worked in a computer store back in the 90's well yea I just better play it safe.



And of course the CPU itself.



I like to add the highest heat that happened when taking my tests with my G. Skill PC3200 Ram. The tests were ran during a heat wave where as my room got very very warm. This was taken when I was Cine Benching this @ 34c. So overall the rig ran pretty cool just using air.



Yea. I am so enjoying the 5900OEM and the additional RAM as it gives me that additional "Ommph" on things I just do not have to worry about.

 

[QuietBob]

So here it is. The goal is to see how much differences there is in wattage performance between two CPU's that suppose to run at the same wattage with the same components being used. With and without driver updates. Secondary was to see the gaming performance between the AMD 3600 and the AMD 5900UM.

Thanks for your detailed analysis! Happy to see you enjoying your new system. If you're concerned with temperatures and power consumption, I'd strongly recommend enabling V-sync in games and lowering your monitor's refresh rate to match. Also, your system can be tweaked to consume much less than 100w in idle.

Sorry to hear about your recent mishap. Stay healthy!