Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

needtoupgrademacpro

macrumors newbie
Original poster
Nov 29, 2018
19
0
Any ideas on why it's maxing out?

I actually tried a speed test on my Hackintosh which goes the full 1 gbps...

I'm running Sierra on my MP 5,1.

Unfortunately, I cannot update to High Sierra on my MP due to critical software compatibility issues.

I have a 860 samsung ssd installed on my MP 5,1.
[doublepost=1544228260][/doublepost]ok... wtf...?

So I just tried the speedtest on chrome vs. safari, and chrome actually shows the full speed....

I'm confused now... are these speed tests optimized for chrome??

I tried the test in the hackintosh via safari and it still showed full 1gbps, as opposed to my MP 5,1 which capped at 400mbit.
 
Last edited:

AidenShaw

macrumors P6
Feb 8, 2003
18,667
4,677
The Peninsula
Any ideas on why it's maxing out?

I actually tried a speed test on my Hackintosh which goes the full 1 gbps...

I'm running Sierra on my MP 5,1.

Unfortunately, I cannot update to High Sierra on my MP due to critical software compatibility issues.

I have a 860 samsung ssd installed on my MP 5,1.
[doublepost=1544228260][/doublepost]ok... wtf...?

So I just tried the speedtest on chrome vs. safari, and chrome actually shows the full speed....

I'm confused now... are these speed tests optimized for chrome??

I tried the test in the hackintosh via safari and it still showed full 1gbps, as opposed to my MP 5,1 which capped at 400mbit.
Your post is missing a lot of specifics that might help us.

What tool are you using to measure the throughput? You mentioned Chrome and Safari - but a web browser typically isn't the best way to do low level hardware benchmarks.

Your hack does 125 MB/s (1000 Gb/s)?

What are the source and sink systems, and what is the network hardware? (It's all copper, right? No WiFi...)
 

jtara

macrumors 68020
Mar 23, 2009
2,008
536
I have a 860 samsung ssd installed on my MP 5,1.

Irrelevant. Your SSD has no bearing on an Internet speed test.
[doublepost=1544232750][/doublepost]
So I just tried the speedtest on chrome vs. safari, and chrome actually shows the full speed....

I'm confused now... are these speed tests optimized for chrome??

Assuming you are referring to Ookla Speedtest, there is a MacOS app available in the Mac App Store. It is the most accurate way to do this sort of test on a Mac.

https://itunes.apple.com/us/app/speedtest-by-ookla/id1153157709?mt=12


[doublepost=1544233002][/doublepost]
I tried the test in the hackintosh via safari and it still showed full 1gbps, as opposed to my MP 5,1 which capped at 400mbit.

Lots of possible reasons. Might be other devices in your home using the Internet at the same time. Your connection might be shared with others (I have a 1Gbps connection through WebPass that is shared with whatever of the 84 other residents of my building happen to be using the Internet - even so, I did a test last night and got 750 down, 900 up). The route between your ISP and the test site you choose might have excessive traffic. It might be a busy time of day. Did you try different test sites?

This is an imperfect test at best. You would only ever see the max if there is NO choke point between your desktop and the test site on the Internet.

The ideal situation would be to test using a test server at your ISP, in their local data center. Anything beyond that, and you are testing the Internet and some random test server, not your connection to the Internet.


[doublepost=1544233050][/doublepost]
What tool are you using to measure the throughput?

From context, I assume Ookla SpeedTest. OP should clarify.

----
Start by testing your local network. Use iperf3. You run a little server on one device, and a test client on another.

There is a kind of test that can be done on WiFi to test the speed between a device and router. CloudCheck is one iOS app that can do this. It has a test called "WiFi SweetSpots" that will make a plot as you walk around your home. It doesn't track your location, but you can stop and make bookmarks. If you have certain Asus routers, you can use a test they call Smartifi as well that is supposed to be a better test. I have an Asus router, but not the right one.
 
Last edited:

needtoupgrademacpro

macrumors newbie
Original poster
Nov 29, 2018
19
0
Your post is missing a lot of specifics that might help us.

What tool are you using to measure the throughput? You mentioned Chrome and Safari - but a web browser typically isn't the best way to do low level hardware benchmarks.

Your hack does 125 MB/s (1000 Gb/s)?

What are the source and sink systems, and what is the network hardware? (It's all copper, right? No WiFi...)

On the MP 5,1 I tried:
fast.com:
Shows 900mbps on chrome, 300-400mbps on safari

speedtest.net:
Same as above...

Speedtest.net app from Mac app store:
Shows 900mbps

On the hackintosh, I tried:
fast.com:
Shows 1gbps both on safari and chrome

speetest.net:
Same as above...

Speedtest.net app from Mac app store:
Shows 1gbps
[doublepost=1544234606][/doublepost]
Irrelevant. Your SSD has no bearing on an Internet speed test.
[doublepost=1544232750][/doublepost]

Assuming you are referring to Ookla Speedtest, there is a MacOS app available in the Mac App Store. It is the most accurate way to do this sort of test on a Mac.

https://itunes.apple.com/us/app/speedtest-by-ookla/id1153157709?mt=12


[doublepost=1544233002][/doublepost]

Lots of possible reasons. Might be other devices in your home using the Internet at the same time. Your connection might be shared with others (I have a 1Gbps connection through WebPass that is shared with whatever of the 84 other residents of my building happen to be using the Internet - even so, I did a test last night and got 750 down, 900 up). The route between your ISP and the test site you choose might have excessive traffic. It might be a busy time of day. Did you try different test sites?

This is an imperfect test at best. You would only ever see the max if there is NO choke point between your desktop and the test site on the Internet.

The ideal situation would be to test using a test server at your ISP, in their local data center. Anything beyond that, and you are testing the Internet and some random test server, not your connection to the Internet.


[doublepost=1544233050][/doublepost]

From context, I assume Ookla SpeedTest. OP should clarify.

----
Start by testing your local network. Use iperf3. You run a little server on one device, and a test client on another.

There is a kind of test that can be done on WiFi to test the speed between a device and router. CloudCheck is one iOS app that can do this. It has a test called "WiFi SweetSpots" that will make a plot as you walk around your home. It doesn't track your location, but you can stop and make bookmarks. If you have certain Asus routers, you can use a test they call Smartifi as well that is supposed to be a better test. I have an Asus router, but not the right one.
See the reply above
 

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
So you’re testing WAN speeds not LAN speeds.
First verify this machine can do it on your LAN (it should be able to) then look at WAN.
Also try dslreports.com/speedtest with the settings for Fiber.
My 3,1 can reach nearly the full link speed on WAN (1Gbps fiber to the house). DSL reports does up to 32 streams which is usually better able to saturate your link. I find it to be more accurate than Speedtest.net the site or the app.

At dsl reports it’s usually in the mid 900Mbps up and down.
[doublepost=1544236701][/doublepost]Use iperf to test between your hack and the Mac Pro. Be sure to test with each machine as the client and the server to simulate traffic in each direction properly. Also test with both of the 1Gb ports on the Mac Pro as maybe you have a bad cable or maybe a bad NIC.
[doublepost=1544236750][/doublepost]Oh and make sure you aren’t accidentally using WiFi.
 
Last edited:

needtoupgrademacpro

macrumors newbie
Original poster
Nov 29, 2018
19
0
So you’re testing WAN speeds not LAN speeds.
First verify this machine can do it on your LAN (it should be able to) then look at WAN.
Also try dslreports.com/speedtest with the settings for Fiber.
My 3,1 can reach nearly the full link speed on WAN (1Gbps fiber to the house). DSL reports does up to 32 streams which is usually better able to saturate your link. I find it to be more accurate than Speedtest.net the site or the app.

At dsl reports it’s usually in the mid 900Mbps up and down.
[doublepost=1544236701][/doublepost]Use iperf to test between your hack and the Mac Pro. Be sure to test with each machine as the client and the server to simulate traffic in each direction properly. Also test with both of the 1Gb ports on the Mac Pro as maybe you have a bad cable or maybe a bad NIC.
[doublepost=1544236750][/doublepost]Oh and make sure you aren’t accidentally using WiFi.
So I tried dslreports.com/speedtest on safari and chrome.

there was a much smoother test when doing this on chrome vs safari.. safari had TONS of bufferbloat

See the attached images

Oh and yes, I was testing this on ethernet, not wifi.
 

Attachments

  • SpeedChrome.png
    SpeedChrome.png
    88.2 KB · Views: 277
  • SpeedSafari.png
    SpeedSafari.png
    87.3 KB · Views: 280

jtara

macrumors 68020
Mar 23, 2009
2,008
536
Looks exemplary to me. You don't have any problem with your "internet".

And either browser speed is way more than needed for human browsing.

You might want to make sure to use Google if you are downloading something big.

Your router is another factor. Do you have any QOS enabled?

I think you're picking nits.

Hadn't used the DSLReports speed test in a long time. I guess they run multiple simultaneous tests to multiple test sites. That's smart! It helps remove route-specific issues, and is "nicer" to the test sites to boot.

Here's what I got from DSLReports. neighbors must be all out shopping and not home streaming...

http://www.dslreports.com/speedtest/42873881
 
Last edited:

jtara

macrumors 68020
Mar 23, 2009
2,008
536
Good detective work!

BTW, both speedtest.net and dslreports.com do offer an http: connection. Not sure what the app uses.

Maybe Chrome has some https: tricks up it's sleeve for processors that don't have the AES instruction set, and uses it's own code?

BTW, from that thread, Geekbench 4 single-thread crypto score was 147 (ouch!) on the W and 1636 on the X.

I just tested iMac Pro at 4408 (29975 multi 8-core/16 thread).

And 2012 Mac Mini i7 at 2770 (8311 multi 4-core/8 thread). 2012 Mac Mini i7 is still in a sweet spot. ;)

But pretty sure the iMac Pro uses the T2 chip for SSL encryption, rather than the CPU, at least for the system framework (and, so, Safari).

Yow, the Nehalam really sucks at crypto!

Interesting to see the huge gains on crypto score, indicating how important it has become and pushed CPU capabilities.
 
Last edited:
  • Like
Reactions: needtoupgrademacpro

needtoupgrademacpro

macrumors newbie
Original poster
Nov 29, 2018
19
0
Good detective work!

BTW, both speedtest.net and dslreports.com do offer an http: connection. Not sure what the app uses.

Maybe Chrome has some https: tricks up it's sleeve for processors that don't have the AES instruction set, and uses it's own code?

BTW, from that thread, Geekbench 4 single-thread crypto score was 147 (ouch!) on the W and 1636 on the X.

I just tested iMac Pro at 4408 (29975 multi 8-core/16 thread).

And 2012 Mac Mini i7 at 2770 (8311 multi 4-core/8 thread). 2012 Mac Mini i7 is still in a sweet spot. ;)

But pretty sure the iMac Pro uses the T2 chip for SSL encryption, rather than the CPU, at least for the system framework (and, so, Safari).

Yow, the Nehalam really sucks at crypto!

Interesting to see the huge gains on crypto score, indicating how important it has become and pushed CPU capabilities.
This is very interesting. A shame Nehalem sucks at encryption.

Thankfully I have my hackintosh for the time being until Apple releases the next modular Mac Pro (let's just wait to see if its decent or not!)

Thanks for the reply friend!
 

needtoupgrademacpro

macrumors newbie
Original poster
Nov 29, 2018
19
0
Looks exemplary to me. You don't have any problem with your "internet".

And either browser speed is way more than needed for human browsing.

I think you're picking nits.

I disagree with this point. Download speeds matter to me since I work remotely and I deal with dropbox and google drive syncing very often, so time and speed is precious to me.

That's the main reason why I upgraded to gigabit service. If you don't get what you pay for, then why pay for it ;)
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.