Because that's NOT testing the GPU at all.
In this case, it's testing the CPU single thread + driver efficiency.
You can install a 1070 and run a network speed test. It also won't improve at all, because it's NOT testing the GPU.
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.
MP All Models Mac Pro CPU Compatibility List
- Thread starter ActionableMango
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
Because that's NOT testing the GPU at all.
In this case, it's testing the CPU single thread + driver efficiency.
You can install a 1070 and run a network speed test. It also won't improve at all, because it's NOT testing the GPU.
Also, because GPU's haven't really gotten a whole lot more "powerful" in the last decade either- but far more efficient at what they do. Think increased stamina over increased strength. Improvements have occurred-but it's more because of their low end processing increasing/lifting the central tendency (avg/min frame rates).
Memory is where the biggest improvements have been made and none of that really matters in benchmarks.
Last edited:
Intel has all the info with the compare function.
X5690 is unlocked.
And it can be OC in Windows (with ThrottleStop) even on cMP. That's proved with benchmarks.
Many years a reader, first time poster.
macnu, I read this post the same day you posted it. I wanted to thank you for doing so. Inspiring! And a year later than you I am in the position where I've outgrown my 2x2.4GHz & 16GB cMP. I'm thinking about going with two X5690 as you did and with either 48 or 96GB of RAM, possibly with a newer GPU also to be able to run Mojave.
May I ask what you do with your Mac, what GPU you installed to be able to run Mojave and how you experience its performance a year after the upgrade?
Also .. Thank you everyone for contributing and keeping this forum as great and informative as it is.
hello. i'm a dual processor mac pro 5,1 newbie so if i say something stupid or that doesn't make sense, let me know. so i'm looking at the Westmere-EP CPUs on wikipedia and i cannot decide on what CPU i want to get. according to my simple minded math, the 6 core CPUs give you the most performance/wattage. based on looking at the TDP, core count, and clock speeds, i've narrowed down my selection to the following: L5640, x5675, x5680. these are also in order from cheapest to most expensive when buying them used on ebay, and also their clock speeds. i've read that classic mac pros have very high wattage use when idle and i don't want to pay the high electricity bill for years to come. when it comes to the x5675 or the x5680, it seems that the x5675 gets better performance/wattage and is slightly cheaper, but what i want to know is how much wattage is being used when the CPUs are idle. also, does web browsing or watching youtube videos count as idle? the question i have about the L5640 is whether this CPU will have significantly lower idle watts than the x5675/x5680 CPUs. 90% of the time, my mac pro will not be in heavy usage but i don't want to compromise on my performance unless if the L5640 has significantly better idle watts.
now that i'm on the subject of watt usage, is there anything else i can do to lower the watts? should i unplug the dvd drive since i never use it? i also plan on upgrading my GT 120 to an RX 560 since i need more GPU power as well. will the RX 560 have lower watt usage when at low level usage?
i know that getting a new computer is an option but i want to hold that off for a few more years
now that i'm on the subject of watt usage, is there anything else i can do to lower the watts? should i unplug the dvd drive since i never use it? i also plan on upgrading my GT 120 to an RX 560 since i need more GPU power as well. will the RX 560 have lower watt usage when at low level usage?
i know that getting a new computer is an option but i want to hold that off for a few more years
Back when I measure mine, it was about 190 watts at idle, and about 30 watts when sleeping. This was for a stock 4,1, single 2.93 4-core CPU, 8GB memory, and 2xGT120. Plus 4 HDDs and a Super Drive. While I can appreciate your desire to get the most performance per watt, I don't think a cMP is a good platform for this. It was shortly after this machine came to market that Intel started to design for efficiency. Along with system vendors, they've made great strides since then.
The RX560 will make the machine use fewer W when web browsing and especially watching YouTube due to accelerating modern graphics and video that use instructions that the older CPU & GPU of the cMP can't support. Maybe you will need to use Chrome browser for accelerated YouTube, but Safari will probably be the most power efficient browser when just browsing.
- Sure, unplug your optical drive.
- Would you be able to unplug any spinning drives and only use SSD(s)?
- Browsing web and watching videos is far from idling.
- You can lower Watt/h by adjusting down the brightness of your monitor to 25%.
- XEON processors are not clocking down as far as desktop CPUs do. This means they never draw very little power. My Mac Pro idles at about 66% W of what it draws at full load. If you can call that idle ...
- My room is heated by electricity. Due to the wattage my MP draws I am able to turn off the heater of this room as long as I keep my Mac Pro running, even though it is below zero °C outside. If you get what I am saying here, I am not using up more power when using my computer than when turning it off - as I would have to turn on an equally power drawing heater to "stay alive". (Or the room would get really cold.)
- Every RAM module draws power. I'd suggest to go with as few as possible for low power consumption.
- Some graphics cards have dual bios. You may be able to re-program one of the BIOSes to underclock to be able to save some more power there.
- A 4GB graphics card may draw less W than a 8GB.
- Where are you located? 230 Volts is more efficient than 110V for the PSU, or so I've read. If you're on 110V, you may be able to save some W by converting to 230V by some efficient equipment. (I've seen miners do this.)
- Be sure to investigate what your OS loads at start up. You may be surprised by how much some software / drivers draw from both RAM and CPU cycles. I once had a fan controller software loaded as standard - until I learnt that it draw 10% of one of my CPUs ... constantly!
- Use wired keyboard and mouse and turn off or even remove the bluetooth hardware.
- Use wired network and remove any wifi hardware.
- Use an external switch to turn off the power when you don't use your computer. A cMP draws a surprising 30W when turned off.
- A 6 core CPU may be more efficient at solving complex tasks compared to a 4 core. But if you're going to watch movies and browse web, keeping power hungry Xeon 6 cores running mostly at idle all the time will most probably draw more power in the end than doing the same with just 4.
- Just though of it: I believe there are some non-XEON options for the cMP too. Did you check them out for power efficiency?
Last edited:
It's going to depend on GPU used, what's in your PCIe slots, SATA drives used, etc-but around 8-10A showing at idle is normal which is ~96-120w per hour, but as you add more peripherals or GPU's your power idle load will increase. GPU's are usually the biggest determining factor of power usage-but even high-end power-hungry cards at idle are usually only 10-15w. The processors like a GPU will only affect the peak power usage, not so much idle usage. Web browsing is mildly above idle, video streaming is definitely not idle-but even more so above 1080p is really starting to make your GPU draw more power. The X5675 will do you fine and save a few watts- (IMO) go with them unless you need all the processing power you can stuff into the machine. I'm very happy with my dual X5675's.
Last edited:
For those of you with the X5690 procs - I managed to find out the maximum temperature this CPU can handle and that's 106 Celsius. I used Intel's processor diagnostic tool on Windows to uncover this info.
My understanding is, that 106C is the max core temperature (reported by the CPU). That's from the Intel official tool, should be very reliable. And we can easily get this reading via iStat etc.
78.5C is the T-case max, which basically means the max temperature between the surface of the IHS and heatsink. Of course, this is the real official limit. However, there is no way to measure this temperature practically.
AFAIK, about 30C difference between this two temperature is very reasonable.
[doublepost=1550253860][/doublepost]Anyway, for others info. W3690's max core is 101C, T-case max is 67.9C. And both these info are from Intel. So, should be reliable.
Last edited:
going to x5690 & 96GB ram will not give the boost that many expect.
I've just gone from a 2.2 6 core to 2 x 3.46 GHz 6-Core Intel Xeon
its not any faster for every day things (or it does not seem so)
I also lost two 16Gb ram boards from the Ebay upgrade kit within the first 2 months...... (possibly solder ball failure)
Keep in mind that many of the things being sold are "ex-server" equipment......
Temp wise i see 30-60 deg c.. and that is with an ambient of 25 deg .c, but i use Istats menu to keep the fans fairly fast.
Also be CLEAR on the maximum ratings found on a data sheet, that is NOT meant for you to run at those specs, according to industry standards those are the ratings MOST statistical failures occur at............
I've just gone from a 2.2 6 core to 2 x 3.46 GHz 6-Core Intel Xeon
its not any faster for every day things (or it does not seem so)
I also lost two 16Gb ram boards from the Ebay upgrade kit within the first 2 months...... (possibly solder ball failure)
Keep in mind that many of the things being sold are "ex-server" equipment......
Temp wise i see 30-60 deg c.. and that is with an ambient of 25 deg .c, but i use Istats menu to keep the fans fairly fast.
Also be CLEAR on the maximum ratings found on a data sheet, that is NOT meant for you to run at those specs, according to industry standards those are the ratings MOST statistical failures occur at............
Last edited:
where do the xeons come from?
many of alibaba are fakes!
I have gone from 2 x 2.97 to 2 x 3.46 and have an increase in performance in almost every application
many of alibaba are fakes!
I have gone from 2 x 2.97 to 2 x 3.46 and have an increase in performance in almost every application
LOL.....
I have spent >25 years in Asia and would NEVER buy CPU from ALIBABA
The CPU actually came from a guy in the USA... they spec as genuine Intel using Intel tools.
I'm well aware of "re-marking" in China , but i have my own sources inside China ... if I ever had the need.
I actually got 3 HP servers from a Chinese "friend" and they were EX "facebook" servers, I know this because of the LOM setup was not erased and all the internal logs were stored inside NVRAM....
These servers were actually containing "5690"'s that were traceable back thru the HP allocation system. and even they did not show an increase in performance when swapped in.
I have spent >25 years in Asia and would NEVER buy CPU from ALIBABA
The CPU actually came from a guy in the USA... they spec as genuine Intel using Intel tools.
I'm well aware of "re-marking" in China , but i have my own sources inside China ... if I ever had the need.
I actually got 3 HP servers from a Chinese "friend" and they were EX "facebook" servers, I know this because of the LOM setup was not erased and all the internal logs were stored inside NVRAM....
These servers were actually containing "5690"'s that were traceable back thru the HP allocation system. and even they did not show an increase in performance when swapped in.
Well I guess it all depends on what a users list of "everyday tasks" consist-most consumers use wouldn't see much of a difference because the clock rate never gets pushed long enough. A processor requires a workload worthy of it to activate turbo-boost, but it's only meant for intermittent burst to sustain heavy prolonged workloads.
Last edited:
I noticed the latest version of UEFITool has been updated so now it gives the names of a lot of the DXE files in english
and i noticed this one, named CPUInitDXE, i wonder if thats worth poking at for either X5687/X5698 support, or transplanting into an XS3,1s Firmware for westmere support?
and i noticed this one, named CPUInitDXE, i wonder if thats worth poking at for either X5687/X5698 support, or transplanting into an XS3,1s Firmware for westmere support?
Stick to 0.25.1 to reconstruct Mac firmwares, anything newer is buggy. Sometimes you add modules and checksums are totally wrong.
Most recent version can't read some checksums, show Fsys and Gaid checksums as 0x00000000h and 0x00000001h