I was going to say the same thing---find a refurbished iMac Pro. Plenty of horsepower for your needs, yet better cooling than any normal iMac configuration you can get.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
i9+Vega Vs i9+580x (Temps and Fan Noise)
- Thread starter Returnoftheimac
- Start date
- Sort by reaction score
Thunderbolt-3 helps future proof your computer with regards to the GPU. You could pick up quite a nice refurbished eGPU in a few years if needed for the price of upgrading now.
I believe that with reports of 128GBs of RAM working with the base i5 (listed as only 32GBs RAM), the ability to pay a shop to swap out the socketed chip, and using a fast external SSD, any 2019 27” iMac is significantly upgradable and would still be a beast after 8-10 years. I’m also thinking the iMac of 2020 will be redesigned with a “helpful“ T2 chip that will eliminate much of our ability to physically upgrade those machines.
The new iMac is a great upgrade but needs some TLC from apple to bring it up to date more. Additionally, now the cost of a top spec i9 iMac has jumped in price so much I’m waiting until the WWDC in June to see what happens regarding the Mac Pro.
iMac Pro is a dead end machine IMO and would much prefer a modular and upgradeable computer with the choice of my own high end monitor (Eizo). I hope the Mac Pro offers processors from regular i9 up to the exotic Xeon chips and bog standard GPU’s or dual high end stuff for those who need it. Being able to upgrade this as we can afford or need would be the only computer Apple ever needs to make.
If however, (as I fear) we hear zero at WWDC I’ll just order the high end iMac and continue to dream of a truly great Mac Pro again
iMac Pro is a dead end machine IMO and would much prefer a modular and upgradeable computer with the choice of my own high end monitor (Eizo). I hope the Mac Pro offers processors from regular i9 up to the exotic Xeon chips and bog standard GPU’s or dual high end stuff for those who need it. Being able to upgrade this as we can afford or need would be the only computer Apple ever needs to make.
If however, (as I fear) we hear zero at WWDC I’ll just order the high end iMac and continue to dream of a truly great Mac Pro again
Last edited:
I haven’t used a PC for many years so most of my Steam games are pretty old. I installed Apex because I’ve played it some on my One X and it’s free and recent. I believe I had AA on because I was doing 1440p but there is a chance I had it turned off because I think I tried it off at 5K since you don’t need it at that resolution but was only getting 20-25fps at 5K, lol. Also I’ve only tested it for like 15 minutes in training mode. I need track down my Razer mouse and keypad because the Magic Mouse is really sensitive on Windows and keeps changing weapons when I barely touch it.
[doublepost=1554061147][/doublepost]
Probably close because I think CPUs make a whole lot less of a difference compared to GPUs nowadays. It might only be a few percent for gaming. But you could always bump a graphical setting or two down to mid-level setting if you aren’t hitting that. Also as I specified above I need to play longer than the tutorial to see if the 60fps is sustained but I need to find my gaming peripherals. I haven’t played PC games for many years and moved to a new house since then.
All this is being really helpful, but you guys are lucky, I still can't make my mind up whether to update the CPU or the GPU
I'm in a pickle
Poster 90 on the other thread was me.
I just noticed they used a Z370 board to benchmark which means they updated its BIOS to support the 9XXX CPUs.
The motherboard manufacturers didn't put a 95W limit on the current Z390 boards, my MSI one didn't.
So I'm not sure if the Z370 that they are using came with a limit and if updating the BIOS removed it.
But I see that they are using Z390 boards if you order from them and they put "4.7-5.0GHz Turbo" next to the 9900K option.
[doublepost=1554076475][/doublepost]
If you're not gaming 50% of the time or more, get the better CPU.
I understand this thread is for the i9, some of you may want to refer to my i5 and 580x thread here:
https://forums.macrumors.com/threads/i5-9600k-benchmarks.2175695/page-2
I ran some benchmarks using Cinebench R20.
https://forums.macrumors.com/threads/i5-9600k-benchmarks.2175695/page-2
I ran some benchmarks using Cinebench R20.
Did a couple of quick tests. When running X-Plane 10, I was getting 70 FPS with the settings that got me 35 FPS on my Late 2013 iMac. My i9 was running at 4.7 Ghz with it open. 90 C on the i9 which then caused the fan to ramp up. I saw 79 C from the Vega 48.
Running the yes command, 1-3 cores saw a 4.7 Ghz speed. With each more core loaded saw a 100 Mhz drop. With 8 cores running at 4.2 Ghz. Now this was only 50% load technically I stopped at 8 yes commands and not 16.
Running Cinebench... Letting it use all 16 threads, it ran between 3.8-3.9 Ghz. Saw 4 Ghz once, but didn't touch it often. Doing another run limiting it to 1 thread saw a speed of 4.7-4.8 Ghz.
Idle temps are 40 C for the i9 and 33 C for the Vega 48.
Running the yes command, 1-3 cores saw a 4.7 Ghz speed. With each more core loaded saw a 100 Mhz drop. With 8 cores running at 4.2 Ghz. Now this was only 50% load technically I stopped at 8 yes commands and not 16.
Running Cinebench... Letting it use all 16 threads, it ran between 3.8-3.9 Ghz. Saw 4 Ghz once, but didn't touch it often. Doing another run limiting it to 1 thread saw a speed of 4.7-4.8 Ghz.
Idle temps are 40 C for the i9 and 33 C for the Vega 48.
Last edited:
Even if you’re gaming only 49% of the time, I’d say you should get the Vega 48.
Hell, even if you’re gaming only 17% of the time, but want that 17% to be quality time, get the better GPU!
(Of course, a less flippant answer would be that you should weigh how much time you expect to spend on CPU-intensive tasks in Photoshop vs how much you might enjoy some high-graphical-quality “AAA” gaming from time to time, taking into consideration the real-world difference in CPU performance between the 6-core i5 and the 8-core i9 and weighing that against the not implausible possibility that a couple of years from now, Photoshop might take a lot more advantage of GPU compute than it does now.)
If one plans to use it as a one-and-done or have disposable funds, max it out to reap its benefits now.
I would only recommend a maxed one to a video person that can't afford a higher spec iMP.
I'd rather use the savings for an eGPU.
If you're patient, down the road there will be options better than Vega and you also take the GPU heat away from the iMac innards.
For the games that can't maintain at least 1440p/60FPS medium settings on the 580X, you can still have a pleasant time locking it to 30FPS.
That’s a good point. Even with the M395X in my 2015 iMac, most games are perfectly playable at 1440p (not always at 60 fps, and not always on ultra-quality settings, of course, and certainly not with the newest “AAA” games), and the 580X is a good deal more powerful than my M395X.
I’m upgrading to a 2019 iMac not because the M95X is hopelessly inadequate, but because the Vega 48 will be a huge upgrade for me, because with Thunderbolt 3 (which the 2015 iMacs don’t have) I can continue to upgrade the GPU over time, and because the 2019 iMacs will be last iMacs that can run all my 32-bit apps and games.
I ran the Shadow of the Tomb Raider benchmark this evening. 1080p on “highest” graphics setting was 56fps average. 1440p on “high” graphics setting was 47fps. Okay but not great, though I suppose Shadow of the Tomb Raider is more of an Nvidia optimized title. I’d say I’m slightly disappointed with the Vega 48 but I shouldn’t be surprised. It’s geared more towards compute and less towards games and it’s inside an all-in-one system. It’s in this weird land between laptop gaming and desktop gaming. New games aren’t going to be buttery smooth at 1440p except sometimes on medium settings and more so if they’re an AMD optimized game. Games from a few years ago are probably going to be fine for the most part. I’d be curious to see what this thing can do with a Vega 56 or 64 eGPU attached but from what I’ve read in the past getting that to work on Windows can be tricky.
I'm planning to get i5 + 980x(stock basically, will only swap hdd for ssd) but a bit worried about temperature, i'm not planning to play AAA, but running old games occasionally would be good, anyone has temperature tests with this combination?
Which score did you get from Cinebench (R20?)?
Could you maybe run the Cinebench R20 benchmark and post the score? You also have the i9+Vega, right?
Thanks guys!
Can't remember. I was focusing on temps and clockspeed. Right now I am getting windows setup on it and currently downloading DCS.
Once that is done, I'll redo cinebench. I will also speed up the fan to max before running it to see if that allows better thermal headroom and can run at a higher clockspeed.
Windows. It's not on the Mac yet AFAIK but should be coming at some point later this year.
Please also do a test without manually speeding up the fan and tell us which power draw and temperature you get from the CPU during the test, and at which RPM they are running, if possible. Thx!
I do not like it. The graphic card performance gain is too small relative to 580. And this is in 2 years ...
With fan manually set to 2600 RPM:
Score: 4197
Temp: 72 C
Speed: 3.9-4.0 Ghz
Fan on Auto:
Score: 4181
Temp: 82 C
Speed: 3.8-3.9 Ghz
So fan set at full blast keeps the CPU cooler and creates a little bit more headroom to maintain a bit faster clock speed
Limiting it to 1 thread made no difference in clock speed or temp with fan going full blast or on auto control.
What I find especially encouraging about these results (which match others I’ve seen online) is that you don’t have to lock your fans to their highest and loudest setting to get what is for all practical purposes the same top performance.
Note that while your clock speed was on average 2.6% higher with your fans manually set to 2600 RPM, your Cinebench R20 score was only 0.38% higher, which is negligible!