There can be a big difference. Those PCIe 3.0 X4 SSD can only negotiate at PCIe 1.0 X4 natively in slot 2 (for 4,1/5,1) without any PCIe switch inside the connection (before the firmware update). If the slot 2 can properly negotiate at PCIe 2.0 X4 with those SSD, their sequential performance will be doubled. And this will make the 4,1/5,1 can use two NVMe in RAID 0 to achieve 3000MB/s by installing one SSD in slot 2 and one SSD in slot 3 with those cheap adaptor. This wasn’t possible before. Because the NVMe will be limited to 750MB/s in slot 2 (without the help of those expensive adaptor e.g. High Point 7101A), and cause the whole SSD RAID 0 limited to 750MB/s x2, which is only 1500MB/. If install the SSD in slot 3 and 4, due to there are only 4 lanes for slot 3 and 4 to share. The ultimate speed is still 1500MB/s. --- Post Merged, Aug 16, 2018 --- I also want to know if there is anything changed. There are four cases. We need at least two cards to done all the checks. 1) flashed card, no loading 2) flashed card, with loading (e.g. CUDA-Z), THEN open system info PCIe Page. 3) unflashed card, no loading 4) unflashed card, with loading (e.g. CUDA-Z), THEN open system info PCIe Page. So far, we know 1 and 3 will shows 2.5GT/s, nothing changed after firmware update. But 2 and 4 still need tests to confirm. --- Post Merged, Aug 16, 2018 --- But stay at 80C with low fan speed seems not quite right. Do you know the fan profile, target temperature etc for the Vega FE?
That was a strange anomaly. There was no reason for the temp to hike like that. The Vega was idle. Anyway I was able to replicate the momentary fan spin down by using the HEVC playback with IINA. It probably has to do with Mojave trying to work better with Polaris/Vega GPUs, which is a good thing if Apple's actually going that direction.
No it's not. At idle the fan should stay at ~760 RPM and the card should hover around 36-42ºC. At least that's the case in Windows. Core should down clock to 500 MHz and HBM2 should be at 167 MHz. Can't remember SoC clock but probably 600 or 800 MHz.
--- Post Merged, Aug 16, 2018 --- You really thought opening Cuda-Z would make a difference? Well, I tried looking at PCI info with both Cuda-Z and OceanWave open (not simultaneously). NO DIFFERENCE! 2.5GT/s. So cross 2) off your list. Lou
Thanks for the test. So, three down, one to go. TBH, no idea, that’s why I prefer we run the test. That system info page often read the real time info at the moment that we open it. So, stress the card before capture the value may make the difference.
Just did the firmware upgrade 138.0.0.0.0 with the Mojave installer. EFI-GTX Titan X (Maxwell) No progress bar, but progress text. Cannot confirm that the firmware upgrade took longer than other firmware upgrades before (about 30 seconds).
Booting from a spare 10.13.6 high Sierra, my 2010 finally made the jump to 0138. Backing up the 0138 rom with romtool, I was greeted by a new warning screen. At least the 2010 is half way there. Any thoughts on which chip is correct for the 2010?
Thanks. So it 'should work'. Looking at the chip to read the number is still suggested. The new version seems to work on Mojave b7.
Never got the multiple definitions screen, flashed a dozens of times this week with v2.0, check your SPI-Flash before trying anything. SST25VF032B, btw.
I got that screen with every flash and used the default profile. I was using v. 2.0 I should have, but wasn’t able at check my chip. dosdude said the default profile for cMP would be correct since using RomTool v.1 presented no problems. It’s the same as v.1 I’ll check the default for mine and report back. Mine is a Mid-2012.
I'll reboot and turn SIP off so I can check what mine defaulted to. BRB... Update: Mine used the same default @handheldgames
Mine was built in Feb 2013, so perhaps by that time, they were using whatever parts were still left for these things. --- Post Merged, Aug 17, 2018 --- Mine reports the same chip. No way to be 100% without looking, but odds are in your favor. Since you flashed with v1 of the tool, there is every reason to believe this will be trouble free for you.
For info, this is NOT the 1st time that Apple provide the GPU audio drive in beta OS. They also did that in High Sierra beta, but then remove that in the official release.
Oh really? I didn't know that. Was it in all the 10.13 betas or just the early ones? Also, do you know which card(s) got it enabled? FWIW it's not enabled for me on the "recommended" Sapphire Pulse RX 580 8GB on bootrom 138 and Mojave PB6. I have submitted a bug report on it though, and would really encourage everyone else with one of the two recommended cards to do so.
Thx for the sharing your Sram configuration. I’ll update the efi on Monday. As you mentioned, it worked the last time I flashed and it should work this time as well. The 2010 is installed in a location that limits accessibility to inner bays for a close up shot of the sram to identify it. --- Post Merged, Aug 17, 2018 --- Thinking about it. I have the HDMI audio driver patch installed. It had been non functional for a while.
Just the very early one, 10.13.0 beta, not on any other 10.13.x beta. I can’t remember which card, I may made a post for that before, but most likely are the R9 380 (if not, then it will be the R9 280). Yeah, please, everyone, submit a bug report for this. --- Post Merged, Aug 17, 2018 --- Then most likely just the new kext allow the HDMIAudio to load properly again, not really has the audio driver included.
HDMI audio driver loaded for Vega up until 10.13.5. The driver AppleGFXHDA.kext is the one that no longer loaded.
FYI, Sapphire Pulse RX580 8GB is $239 right now at Newegg. Plus, you can get an additional $25 off if you check out with Masterpass. It’s so nice to see these cards coming back down to their proper pricing.