Description of problem:
I first observed that my server had 20% slower performance with Hyperthreading turned on.  It is a 2P Intel E5-2680 v3 with DDR4 memory.  Checking logs showed a recurring crash in all Fedora 20 and 21 kernels to date.  I cannot tell if this is an OS bug or a SuperMicro bios problem.  The motherboard is a Supermicro X10DRi, BIOS 1.0b dated 09/17/2014.  Dmesg reports the bios is "broken" and warns that performance will be degraded.  I also filed this bug with Fedora 20 just to document its presence (see bug 1185660 at Red Hat Bugzilla).  I plan to follow developments in this bug report and plan to ask for assistance from SuperMicro.

Reproducing it is easy - just boot the system and ABRT reports a kernel crash and recovery.  It consistently happens each time the system is booted, with or without Hyperthreading (24 or 48 cpus).

Additional info:
reporter:       libreport-2.3.0
WARNING: CPU: 0 PID: 4 at drivers/dma/ioat/dca.c:697 ioat3_dca_init+0x192/0x1af [ioatdma]()
ioatdma 0000:00:04.0: APICID_TAG_MAP set incorrectly by BIOS, disabling DCA
Modules linked in: ioatdma(+) wmi(+) acpi_power_meter(+) ipmi_si tpm_tis ipmi_msghandler tpm lpc_ich mei_me sb_edac mei acpi_pad edac_core i2c_i801 mfd_core shpchp nfsd auth_rpcgss nfs_acl lockd sunrpc uas usb_storage ast drm_kms_helper ttm drm igb ptp pps_core dca i2c_algo_bit
CPU: 0 PID: 4 Comm: kworker/0:0 Not tainted 3.17.8-300.fc21.x86_64 #1
Hardware name: Supermicro X10DRi/X10DRi, BIOS 1.0b 09/17/2014
Workqueue: events work_for_cpu_fn
 0000000000000000 0000000084f1b57a ffff8808542d7c68 ffffffff81741aea
 ffff8808542d7cb0 ffff8808542d7ca0 ffffffff810970bd ffff880853315000
 ffff8808532f83b0 ffff880845234c80 ffffc90016c50100 0000000000000100
Call Trace:
 [<ffffffff81741aea>] dump_stack+0x45/0x56
 [<ffffffff810970bd>] warn_slowpath_common+0x7d/0xa0
 [<ffffffff810971b4>] warn_slowpath_fmt_taint+0x54/0x70
 [<ffffffffa021bb42>] ioat3_dca_init+0x192/0x1af [ioatdma]
 [<ffffffffa021aebe>] ioat3_dma_probe+0x35e/0x450 [ioatdma]
 [<ffffffffa021318e>] ioat_pci_probe+0x11e/0x1b0 [ioatdma]
 [<ffffffff813c6ac5>] local_pci_probe+0x45/0xa0
 [<ffffffff810ac704>] work_for_cpu_fn+0x14/0x20
 [<ffffffff810af89d>] process_one_work+0x14d/0x400
 [<ffffffff810b038c>] worker_thread+0x1cc/0x4a0
 [<ffffffff810b01c0>] ? rescuer_thread+0x2a0/0x2a0
 [<ffffffff810b531a>] kthread+0xea/0x100
 [<ffffffff810b5230>] ? kthread_create_on_node+0x1a0/0x1a0
 [<ffffffff81748bfc>] ret_from_fork+0x7c/0xb0
 [<ffffffff810b5230>] ? kthread_create_on_node+0x1a0/0x1a0

Potential duplicate: bug 1185660