Can't get hashcat 3.x to use my Radeon 5970

I've tried Radeon Crimson drivers (16.2.1 and beta), Radeon Catalyst 15.7 and Radeon Catalyst 14.9. I've tried installing OpenCL™ Runtime for Intel® Core™ and Intel® Xeon® Processors 16.1. 

oclHashcat 2.01 works great at ~160KH/s.  With Hashcat 3.10 it seems to detect my card but I can't ever get over 3000 H/s no matter which combination of platform (--opencl-platforms) and opencl device settings I use. Can anyone help me out with this? Thanks.

Hashcat64.exe
-----------
C:\Users\user\Downloads\hashcat-3.10\hashcat-3.10>Hashcat64.exe  --opencl-platfo
rms=1 -d 1,2 -m 2500 -a3 123.hccap ?d?d?d?d?d?d?d?d?d
hashcat (v3.10) starting...

OpenCL Platform #1: Advanced Micro Devices, Inc.
================================================
- Device #1: Cypress, 512/1024 MB allocatable, 20MCU
- Device #2: Cypress, 512/1024 MB allocatable, 20MCU
- Device #3: WARNING: Not a native Intel OpenCL runtime, expect massive speed loss
             You can use --force to override this but do not post error reports if you do so
- Device #3:        Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz, skipped

OpenCL Platform #2: Intel(R) Corporation, skipped

Hashes: 1 hashes; 1 unique digests, 1 unique salts
Bitmaps: 16 bits, 65536 entries, 0x0000ffff mask, 262144 bytes, 5/13 rotates
Applicable Optimizers:
* Zero-Byte
* Single-Hash
* Single-Salt
* Brute-Force
* Slow-Hash-SIMD
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 75c

- Device #1: Kernel m02500.e3d4061c.kernel not found in cache! Building may take a while...
- Device #1: Kernel markov_le.e3d4061c.kernel not found in cache! Building may take a while...

- Device #1: Kernel amp_a3.e3d4061c.kernel not found in cache! Building may take a while...

[s]tatus [p]ause [r]esume ypass [c]heckpoint [q]uit =>

Session.Name...: hashcat
Status.........: Running
Input.Mode.....: Mask (?d?d?d?d?d?d?d?d?d) [9]
Hash.Target....: 123 (xxxxxx:99:3f:0a <-> xxxxxx:12:58:5b)
Hash.Type......: WPA/WPA2
Time.Started...: Thu Oct 06 23:19:12 2016 (2 secs)
Time.Estimated.: Fri Oct 07 00:41:15 2016 (1 hour, 21 mins)
Speed.Dev.#1...:     1336 H/s (13.96ms)
Speed.Dev.#2...:      699 H/s (12.74ms)
Speed.Dev.#*...:     [b]2035 H/s
Recovered......: 0/1 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 0/10000000 (0.00%)
Rejected.......: 0/0 (0.00%)
Restore.Point..: 0/10000000 (0.00%)
HWMon.Dev.#1...: Temp: 52c Fan: 29% Util:  0% Core: 725Mhz Mem:1000Mhz Lanes:16
HWMon.Dev.#2...: Temp: 51c Util:  0% Core: 725Mhz Mem:1000Mhz Lanes:16

-----------

oclhashcat64.exe
-----------
C:\Users\user\Downloads\oclHashcat-2.01\oclHashcat-2.01>oclHashcat64.exe -m 2500
 -a3 123.hccap ?d?d?d?d?d?d?d
oclHashcat v2.01 starting...

Device #1: Cypress, 1024MB, 725Mhz, 20MCU
Device #2: Cypress, 1024MB, 725Mhz, 20MCU

Hashes: 1 hashes; 1 unique digests, 1 unique salts
Bitmaps: 16 bits, 65536 entries, 0x0000ffff mask, 262144 bytes, 5/13 rotates
Applicable Optimizers:
* Zero-Byte
* Single-Hash
* Single-Salt
* Brute-Force
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 80c
Device #1: Kernel ./kernels/4098/m02500.Cypress_1573.4_1573.4 (VM)_1449318458.ke
rnel (540088 bytes)
Device #1: Kernel ./kernels/4098/markov_le_v2.Cypress_1573.4_1573.4 (VM)_1449318
458.kernel (51704 bytes)
Device #1: Kernel ./kernels/4098/amp_a3_v2.Cypress_1573.4_1573.4 (VM)_1449318458
.kernel (16028 bytes)
Device #2: Kernel ./kernels/4098/m02500.Cypress_1573.4_1573.4 (VM)_1449318458.ke
rnel (540088 bytes)
Device #2: Kernel ./kernels/4098/markov_le_v2.Cypress_1573.4_1573.4 (VM)_1449318
458.kernel (51704 bytes)
Device #2: Kernel ./kernels/4098/amp_a3_v2.Cypress_1573.4_1573.4 (VM)_1449318458
.kernel (16028 bytes)

Input.Mode.....: Mask (?d?d?d?d?d?d?d?d?d) [9]
Hash.Target....: 123 (xxxxxx:12:58:5b <-> xxxxxxx:99:3f:0a)
Hash.Type......: WPA/WPA2
Speed.GPU.#1...:    79375 H/s
Speed.GPU.#2...:    79411 H/s
Speed.GPU.#*...:   158.8 kH/s
-----------

[/b]